ALG13 participates in epileptogenesis via regulation of GABAA receptors in mouse models

ALG13 (asparagine-linked glycosylation 13) plays crucial roles in the process of N-linked glycosylation. Mutations of the ALG13 gene underlie congenital disorders of glycosylation type I (CDG-I), a rare human genetic disorder with defective glycosylation. Epilepsy is commonly observed in congenital disorders of glycosylation type I (CDG-I). In our study, we found that about 20% of adult ALG13KO knockout mice display spontaneous seizures, which were identified in a simultaneous video and intracranial EEG recording. However, the mechanisms of ALG13 by which deficiency leads to epilepsy are unknown. Whole-cell patch-clamp recordings demonstrated that ALG13KO mice show a marked decrease in gamma-aminobutyric acid A receptor (GABAAR)-mediated inhibitory synaptic transmission. Furthermore, treatment with low-dose diazepam (a positive allosteric modulator of GABAA receptors), which enhances GABAAR function, also markedly ameliorates severity of epileptic seizures in ALG13KO mice. Moreover, ALG13 may influenced the expression of GABAARα2 membrane and total protein by changing transcription level of GABAARα2. Furthermore, protein interactions between ALG13 and GABAARα2 were observed in the cortex of wild-type mice. Overall, these results reveal that ALG13 may be involved in the occurrence of epilepsy through the regulation of GABAAR function, and may provide new insight into epilepsy prevention and treatment.


Introduction
Epilepsy is one of the most common neurological disorders that affects 1-2% of the population worldwide across different ages and background. Epilepsies are characterized by recurrent seizures and are caused by the imbalance of excitatory and inhibitory neural circuits 1,2 . Asparagine-linked glycosylation 13 (ALG13) is located on the X-chromosome and encodes a protein that heterodimerizes with ALG14 to form a functional UDP-GlcNAc glycosyltransferase in the endoplasmic reticulum that catalyzes the second step of protein N-glycosylation 3,4 . Protein asparagine N-glycosylation is thought to be essential for the structure and function of glycoproteins 5 . Previous reports implicated ALG13 gene mutations were regarded as a cause of X-linked congenital disorders of glycosylation type I (CDG-I), also known as ALG13-CDG 6,7 . CDG-I, a rare human genetic disorder, characterized by defective glycosylation pathways. Asparagine N-glycosylation is crucial for multiple biological processes and thus CDG patients present highly variable multisystem phenotype that including developmental delay/ intellectual disability, muscle hypotonia, seizures, endocrine and coagulation abnormalities, and psychomotor retardation in multiple organ systems [8][9][10][11] . Epilepsy is commonly observed in CDG and is often a presenting symptom 8 . In research by Barba et al. 12 , partial seizures were observed in 4 of 17 consecutively observed children with CDG (three females, one male), related testing and analysis indicated CDG-I, two patients were proved with homozygous or compound heterozygous variants through genetic testing. Similarly, in study by Timal et al. 13 , a missense mutation in the X-linked gene ALG13 was found in a child with epilepsy and early death caused by CDG-1.
Previous studies in our laboratory have reported that a wide distribution of ALG13 was observed in the mice brain, particularly in cortex and hippocampus, which is the primary zone closely related to epilepsy 14 . Moreover, our study also shows that ALG13 deficiency increased seizure susceptibility induced by kainic acid or pilocarpine 14 . To our surprise, about 20% of ALG13KO mice with frequent epileptiform convulsion were observed accidentally later. In light of this direct evidence, we intend to study the role of ALG13 in epileptic seizures.

ALG13KO mice display spontaneous seizures
There were no obvious difference in the behavior and appearance of ALG13KO mice and wild-type littermates at birth. Around 20% of adult ALG13KO mice began to exhibit frequent spontaneous seizures, female mice particularly mostly, especially during cage changing and handling. Typically, seizures consisted of a behavioral sequence including (i) movement arrest; (ii) repeated clonic jerks, restricted to forelimbs, sometimes unilaterally, followed by a hypertonic neck and tail, and accompanied by loss of posture; and (iii) uncontrolled jumping and running, frequently incontinence and loss of postural equilibrium (Fig. 1). Seizures usually ended after hypertonic postures, during the postictal period, mice were immobile for several seconds. Wild-type littermates never displayed spontaneous epileptic manifestations.

Video-EEG recording demonstrates epileptic activity in ALG13KO mice
To explore the phenotype caused by ALG13 mutation, adult ALG13KO mice were studied in simultaneous video and intracranial cortical electroencephalography (EEG) recordings. Ictal epileptic EEG abnormalities were obvious in ALG13KO mice with spontaneous behavioral seizures, and spontaneous seizures were recorded (Fig.  2a). Typically, low-amplitude fast activities; followed by bursts of polyspikes of increasing amplitude and decreasing frequency (Fig. 2b). Moreover, "EEG seizures," a defined seizure, were evident in ALG13KO mice without spontaneous behavioral seizures (Fig. 3a, b). Epileptic activity was never detected in age-matched wild-type littermates.
In order to determine whether the effect of ALG13 on mIPSCs involves the effect on glycine current, a set of controlled experiments was performed. The results showed that ALG13 deficiency did not change amplitude of mIPSCs after strychnine was added (Fig. 4e, f). Therefore, ALG13 changes GABAergic transmission without affecting glycine-ergic transmission.

Treatment of epileptic phenotypes and EEG abnormalities in ALG13KO mice with diazepam
Given that the epileptic phenotype and EEG abnormalities in ALG13KO mice may emerge from decreased GABAergic inhibitory transmission, we predicted that they could be rescued by enhancing the strength of GABAergic transmission. To test this idea, we tested the effects of diazepam, a positive allosteric modulator of the GABA A receptor, on GABAergic inhibitory transmission in ALG13KO mice cortical slices. As expected, treatment with 0.3-mM diazepam increased mIPSC amplitude (***P < 0.001) (Fig. 4g, h), then we examined the responsiveness of ALG13KO and wild-type mice to diazepam. As expected, both ALG13KO and wild-type mice showed a reduction in epileptic activity following treatment with diazepam (2 mg/kg) as evidenced by a decreased frequency of the "EEG seizure" in the EEG recordings. Due to high doses of diazepam cause sedation, the lower dose of diazepam also was used. We found that increased frequency of the "EEG seizure" and abnormally elevated δ and θ power in ALG13KO mice, which could be alleviated by treatment with diazepam (0.25 mg/kg) (***P < 0.001), but not in wild-type mice (Fig. 5a, b). These results support our hypothesis that ameliorative epileptic phenotype and EEG abnormalities by treatment with diazepam are related to increased strength of inhibitory transmission.

ALG13 influences epileptic activity by reducing GABA A R α2 subunit gene expression
To further investigate the mechanism by which ALG13 affects the inhibitory postsynaptic current in epilepsy, we examined the effect of ALG13 on GABA A R regulation. There was no difference in the expression level of total and membrane GABA A R α1, α5, and γ2 protein between the ALG13KO group and the WT group ( Fig. 6a, b). However, compared with the wild-type group (total: 1.007 ± 0.021; membrane: 0.6108 ± 0.009), the expression of GABA A Rα2 total and membrane protein was reduced in the ALG13KO group (total: 0.8243 ± 0.019; membrane: 0.4350 ± 0.013) (***P < 0.001) (Fig. 6a, b). Furthermore, compared with the wild-type group (1.027 ± 0.018), the expression of GABA A Rα2 mRNA was significantly reduced in the ALG13KO group (0.4267 ± 0.019) (***P < 0.001) (Fig. 7a). Immunofluorescent result showed probable protein-protein interactions between ALG13 and GABA A Rα2 (Fig. 7b) in the cortex of wild-type mice. Taken together, these results suggest that ALG13 affected mRNA expression of Gabrα2 gene, which affected total and surface protein expression of GABA A Rα2 in the cortex. ALG13 might affect the expression of membrane protein via interaction with GABA A Rα2 or by altering the transcription of GABA A Rα2, and further influence mIPSC.

Discussion
We have characterized the phenotype of ALG13KO mice, showing spontaneous seizures and electroclinical characterization of seizures with video-EEG monitoring and providing details of seizure semiology. Furthermore, we found that loss of ALG13 may reduce the inhibitory  Total number of spontaneous "EEG seizure" events from wild-type and ALG13KO mice without any external intervention (EEGs were recorded for 2 weeks). a "EEG seizure" events were detected in C57BL/6J mice implanted with cortical electrodes. Numbers in the first column indicate animal identifiers. Subsequent columns indicate the day during the recording period. Each individual box is color coded according to the legend. Each of the numbers in these boxes indicates the total number of events per day. b Example of an event, defined as a high amplitude (>3 times baseline) rhythmic discharge, frequency (>5 Hz), and lasting longer than 5 s. Bar, 1 mV, 1 s. c Representative traces of EEG activity in WT and ALG13KO mice without any external intervention. Bar, 1 mV, 15 min. d Spectrograms of representative EEG recordings from wild-type and ALG13KO mice during baseline. e FFT of EEG recordings comparing ALG13KO and wild-type mice (***P < 0.001, two-way ANOVA test, n = 6)). f Comparison of average δ and θ power between wild-type and ALG13KO mice (***P < 0.001, two-sided unpaired T-test, n = 6).
synaptic transmission by regulating the transcription of GABA A R α2 subunit.
Our previous studies have investigated the distribution and localization of ALG13 in mouse brain, the histological results suggested that ALG13 was highly expressed in the cortex and hippocampus 14 , which are the primary regions of epileptic pathological alterations 15,16 . Immunofluorescent results showed ALG13 to be expressed in Fig. 4 Results of patch-clamp recording. a Representative traces of mEPSCs in the cortical pyramidal neurons from p21-30 ALG13KO and wild-type littermate control mice. Bar, 20 pA, 2 s. b The amplitude (P > 0.05, two-sided unpaired T-test, n = 6) and the frequency (P > 0.05, unpaired t-test, n = 6) of mEPSCs in the WT and ALG13KO groups. c Representative traces of mIPSCs in the cortical pyramidal neurons from p21-30 ALG13KO and wild-type littermate control mice. Bar, 20 pA, 2 s. d The amplitude (***P < 0.001, unpaired t-test, n = 6) and the frequency (P > 0.05, two-sided unpaired T-test, n = 6) of mIPSCs in the WT and ALG13KO groups. e, f There was not change amplitude of mIPSCs (with strychnine) between ALG13KO group and WT group (n = 6). Bar, 20 pA, 2 s. g, h Treatment with 0.3-mM diazepam increased the amplitude (***P < 0.001, two-sided unpaired T-test, n = 6) of mIPSC in ALG13KO cortical pyramidal neurons. Bar, 20 pA, 2 s. neurons but not in astrocytes. These characteristics of ALG13 expression indicates that ALG13 may be involved in epilepsy.
Video-EEG studies on ALG13KO mice confirmed that seizures were often initiated by behavioral immobility, followed by forelimb licking, clonus, succeeding seizures tended to terminate with wild running and tonic-clonic movements. The epileptic phenotype in these animals is similar to postictal behaviors following chemoconvulsant TLE models in animals 17,18 . Interictal EEG activity, with varied discharges initiating as low-voltage fast activities followed by spike and polyspike-wave discharges of increasing amplitude and decreasing frequency, is similar to intracranial EEG recordings of chemoconvulsant TLE models [19][20][21] . Furthermore, neuropathological changes were found in our previous studies 14 . These changes, which include neuronal cell death, reactive astrogliosis, and aberrant mossy fiber sprouting in the dentate gyrus are typical for human temporal lobe seizures and many animal models of hippocampal seizures [22][23][24] . Taken together, the results above appear to suggest that ALG13KO mice may be an ideal animal model of temporal lobe epilepsy.
The normal functioning of the cortex depends on a perfect balance between excitatory and inhibitory input, so any disturbance of this balance brings the possibility of uncontrolled hyperexcitability 25,26 . In the present study, deficiency of ALG13 decreased mIPSC amplitude. In general, the amplitude of mIPSC discharge reflects the postsynaptic response to released neurotransmitter, and frequency changes are associated with a presynaptic effect 27 . These results suggest that ALG13 may affect the inhibitory synaptic transmission via postsynaptic receptors.
GABA and glycine are the major inhibitory neurotransmitters in the CNS. Glycine-ergic currents very similar to GABAergic, and GABA A R antagonists also act at glycine receptor 28 . Strychnine is an antagonist of Fig. 5 Total number of spontaneous "EEG seizure" events and EEG δ and θ power from WT and ALG13KO mice following diazepam treatment (2 and 0.25 mg/kg). a "EEG seizure" events were detected in ALG13KO mice implanted with cortical electrodes. Numbers in the first column indicate animal identifiers. Subsequent columns indicate the day during the recording period. Each individual box is color coded according to the legend. Each of the numbers in these boxes indicates the total number of events per day (n = 6 per group). Example of an event, defined as a high amplitude (>3 times baseline) rhythmic discharge, frequency (>5 Hz), and lasting longer than 5 s. Average number of spontaneous "EEG seizure" events from WT and ALG13KO mice following diazepam treatment (***P < 0.001, treatment of 2-mg/kg diazepam: WT compared with ALG13KO, oneway ANOVA test, n = 6; ***P < 0.001, treatment of 0.25-mg/kg diazepam: WT compared with ALG13KO, one-way ANOVA test, n = 6) bar, 1 mV, 1 s. b Comparison of baseline average δ and θ power between wild-type and ALG13KO mice and after treatment with 2 and 0.25-mg/kg diazepam (***P < 0.001, baseline: WT compared with ALG13KO, n = 6; **p < 0.01, WT: baseline compared with diazepam (2 mg/kg), n = 6; p > 0.05, baseline compared with diazepam (0.25 mg/kg), n = 6; ***P < 0.001, ALG13KO: baseline compared with diazepam (2 mg/kg), n = 6; ***P < 0.001, baseline compared with diazepam (0.25 mg/kg), n = 6. One-way ANOVA with a post hoc Bonferroni tests was performed.
glycine, which can selectively inhibits glycine-ergic activity. Our results indicated that ALG13 deficiency did not change amplitude from the first (no strychnine) and second (with strychnine) experiment. Thus, we speculated that ALG13 might influence mIPSCs by regulating postsynaptic GABA A Rs.
Benzodiazepine has been widely used to alleviate epileptic seizure 29,30 . Experimental evidence suggested that benzodiazepine can reverse decreased GABAergic tone 31,32 . Our results indicate that treatment with low nonsedating dose benzodiazepine increase inhibitory neurotransmission, which could be a potential Fig. 6 ALG13 affects the total and membrane protein expression of GABA A Ra2. a Representative images of the total expression levels of GABA A Rα1 α2, α5, and γ2 and statistical analyses of the total expression levels of GABA A Rα1, α2, α5, and γ2 determined using western blotting analyses (***P < 0.001, expression levels of GABA A Rα2: WT compared with ALG13KO, two-sided unpaired T-test, n = 6). There were no significant differences in the level of total GABA A R α1, α5, and γ2 between the WT group and the ALG13KO group. b Representative images of the membrane expression levels of GABA A Rα1, α2, α5, and γ2 and statistical analyses of the membrane expression levels of GABA A Rα1, α2, α5, and γ2 determined using western blotting analyses (***P < 0.001, expression levels of GABA A Rα2: WT compared with A LG13KO, two-sided unpaired T-test, n = 6). There were no significant differences in the level of membrane GABA A R α1, α5, and γ2 between the WT group and the ALG13KO group. Fig. 7 ALG13 affects the mRNA expression of Gabrα2; the cell localization of ALG13 and GABA A Rα2. a The expression levels of Gabrα2 mRNA was significantly lower in the ALG13KO group than that in the corresponding control (***P < 0.001, expression levels of Gabrα2: WT compared with ALG13KO, two-sided unpaired T-test, n = 6). There were no significant differences in the level of total Gabrα1, α5, and γ2 mRNA between the WT group and the ALG13KO group. b Immunofluorescent triple-staining of ALG13 (green), GABA A Rα2 (red), and DAPI (blue) in the cortex, scale bar = 20 µm.
pharmacological intervention for patients with mutation in the ALG13.
Many GABA A receptors are composed of two α subunits, two β subunits and one γ subunit. The binding site for benzodiazepines is made up with one of the α subunits (α1, α2, α3, or α5) and a γ subunit (typically the γ2 subunit) 33,34 . The number of GABA A receptors in the postsynaptic membrane directly controls the efficacy of GABAergic synaptic transmission 35 . We investigated the expression of the GABA A Rs in ALG13KO and wild-type mice. There was no difference in the expression level of total or surface GABA A Rα1, α5, and γ2 protein in the cortex between the ALG13KO groups and the WT groups. However, deficiency of ALG13 reduced the expression of GABA A Rα2 membrane protein. Meanwhile, we find that total GABA A Rα2 protein displayed the same change in the cortex. This finding suggests that the synthesis of GABA A Rα2 protein may change. To further test this hypothesis, we used real-time PCR to analyze transcription of mRNA. The results indicated that deficiency of ALG13 affected the transcription level of GABA A Rα2. In addition, In our study, Immunofluorescent revealed that ALG13 interacts with GABA A Rα2.
In conclusion, our findings demonstrate that mutations in ALG13 can led to epilepsy and alleviate GABA A Rmediated postsynaptic transmission. Further, ALG13 regulated the expression of GABA A Rα2 total and membrane proteins possibly through affecting the transcription of GABA A Rα2 or via interaction with GABA A Rα2. These may be major contributor to epileptogenesis. However, the mechanism whereby ALG13 regulates the GABA A receptor still needs to be explored further and in greater detail.

Experimental animals
ALG13 KO mice created with a C57BL/6J background were provided by Y.B. (Shanghai Jiaotong University, Shanghai, China). Gene knockout mice were generated by deleting five nucleotides of the fourth exon of the ALG13 gene using CRISPR-Cas9 systems. Genotyping of ALG13KO mice as previously described in our study (Gao et al. 14 ). All mice were kept in a specific-pathogen-free environment under standard conditions with a 12-h light/ dark cycle and controlled temperature at 22°C with food and water available ad libitum. Animals were randomly assigned to different experimental groups. 10-12-weekold male mice were used in all experiments except the patch-clamp experiment. All procedures used on mice were approved by the Commission of Ningxia Medical University for Ethics of Experiments on Animals and were in accordance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals.

Electroencephalography (EEG)
Surgery and recording 10-12-week-old male mice were used for EEG studies. Prefabricated light-weight EEG electrodes (2-channel) were surgically implanted under isofluorane anesthesia (3-5% for induction, 1-2% for maintenance), placing the electrodes leads into the subdural space. Electrodes were anchored to the skull and over the left frontal cortex (reference electrode) and the left and right parietal cortices as described using stainless steel screws and acrylic cement. All EEG recordings were carried out at least 1 week after surgery on freely moving mice in a recording chamber. Digital EEG activity and videos of their locomotor activity were recorded with SleepSign v.3.0 (Kissei Comtec, Japan). A spontaneous "EEG seizure" event was defined as a high amplitude (>3 times baseline) rhythmic discharge, frequency (>5 Hz), and lasting longer than 5 s 36 . Spectral power was obtained by subjecting the recordings to a fast-Fourier transform using SleepSign. Results were quantified using SleepSign and Matlab.

Diazepam intervention
Four sets of identical trials were performed at 1-week intervals with the same groups of mice. In the first trial, we performed the long-term continuous EEG monitoring without any treatment. In a subsequent trial, the same tests were performed after intraperitoneal injection of 2mg/kg clonazepam. In the third trial, the same tests were performed after intraperitoneal injection of 0.25-mg/kg clonazepam. In the last trial, the tests were performed after intraperitoneal injection of vehicle.
To determine the ability of ALG13 in altering the GABAergic synaptic current or glycine-ergic current, strychnine (0.5 mM, MCE) was infused into ACSF to selectively inhibit glycine-ergic activity. mIPSCs were recorded in the pyramidal neurons after strychnine administration.
After the baseline data were recorded, diazepam (0.3 mM, sigma) was added to the perfusate. All recordings were conducted in standard ACSF with 0.1% DMSO. Signals were obtained from Patch-Clamp Amplifier (HEKA EPC10 double USB, German), filtered at 2 kHz and digitized at 10 kHz, recorded by PatchMaster (HEKA, German). The data were discarded if the series resistance fluctuated by more than 25% of the initial value. Mini Analysis 6.0.1 was used to analyze the recorded trace. Origin Pro 8.0 software was applied to convert exported data to the graphical format.

Western blotting
The cortex was excised, weighed, and washed in phosphate-buffered saline and subsequently homogenized on ice using a Bullet Blender (Next Advance, Inc., Troy, NY, USA) in lysis buffer consisting of phosphatase inhibitors, protease inhibitor, and phenylmethylsulfonyl fluoride. Then, the homogenates were centrifuged at 12000 g for 15 min at 4°C, and the total protein was collected with the supernatant. The membrane proteins were extracted according to the instructions for the Membrane Protein Extraction Kit (KeyGen, Nanjing, China). The homogenate was incubated at 4°C for 10 min and centrifuged at 12,000 rpm at 4°C for 5 min. The precipitate obtained by centrifugation was taken, 200 μl of cold extraction buffer was added, vortexed, and mixed for 30 s, and then placed on ice for 5 min, repeated five times. The homogenate was centrifuged at 12,000 rpm at 4°C for 10 min, and supernatant fractions were collected. The protein concentrations were then determined using a bicinchoninic acid kit (BCA kit; KeyGen, Nanjing, China).

Real-time reverse transcription polymerase chain reaction (RT-PCR)
The total RNA of the cortices was isolated with the TRIzol solution (100-mg tissue/ml; Invitrogen, Carlsbad, CA, USA), according to the manufacturer's protocol. The OD260/280 value of extracted RNA sample was detected utilizing a spectrophotometer. Next, cDNA was synthetized using Revert Aid First Strand cDNA Synthesis Kit (Thermo Scientific, K1621). The gene sequences of the GABAAR subunits were designed from the National Center for Biotechnology Information database.

Experimental design and statistical analysis
A sample size of 6 was arbitrarily chosen for first EEG and whole-cell patch-clamp study. Because there was a statistically significant difference using n = 6, we chose to continue using this sample size for all our studies. According to whether the samples exhibited normal distributions and equal variances (determined by the onesample Kolmogorov-Smirnov test and Levene's test), the experimental results were statistically assessed using parametric or nonparametric tests. All data are expressed as mean ± SD, and comparisons between two groups were performed using unpaired Student's two-tailed t test. One-way analysis of variance (ANOVA) with a post hoc Bonferroni test was used to measure differences between multiple groups considering one fixed factor. Two-way ANOVA with a post hoc Bonferroni test was used to measure differences between two groups comparisons considering two fixed factors. P value < 0.05 was considered to be statistically significant. SPSS 23.0 and GraphPad Prism 5.0 software were used for statistical analyses and graphing, respectively.