Distinct Functional Alterations and Therapeutic Options of Two Pathological De Novo Variants of the T292 Residue of GABRA1 Identified in Children with Epileptic Encephalopathy and Neurodevelopmental Disorders

Mutations of GABAAR have reportedly led to epileptic encephalopathy and neurodevelopmental disorders. We have identified a novel de novo T292S missense variant of GABRA1 from a pediatric patient with grievous global developmental delay but without obvious epileptic activity. This mutation coincidentally occurs at the same residue as that of a previously reported GABRA1 variant T292I identified from a pediatric patient with severe epilepsy. The distinct phenotypes of these two patients prompted us to compare the impacts of the two mutants on the receptor function and to search for suitable therapeutics. In this study, we used biochemical techniques and patch-clamp recordings in HEK293 cells overexpressing either wild-type or mutated rat recombinant GABAARs. We found that the α1T292S variant significantly increased GABA-evoked whole-cell currents, shifting the dose–response curve to the left without altering the maximal response. In contrast, the α1T292I variant significantly reduced GABA-evoked currents, shifting the dose–response curve to the right with a severely diminished maximum response. Single-channel recordings further revealed that the α1T292S variant increased, while the α1T292I variant decreased the GABAAR single-channel open time and open probability. Importantly, we found that the T292S mutation-induced increase in GABAAR function could be fully normalized by the negative GABAAR modulator thiocolchicoside, whereas the T292I mutation-induced impairment of GABAAR function was largely rescued with a combination of the GABAAR positive modulators diazepam and verapamil. Our study demonstrated that α1T292 is a critical residue for controlling GABAAR channel gating, and mutations at this residue may produce opposite impacts on the function of the receptors. Thus, the present work highlights the importance of functionally characterizing each individual GABAAR mutation for ensuring precision medicine.


Introduction
Epilepsy is a neurological disorder characterized by recurrent seizures that occur due to abnormally excessive electrical discharges of the cerebral neurons as a result of the disruption of the excitation-inhibition balance (E/I balance) in the brain [1][2][3]. About 50% of the epilepsies diagnosed worldwide are gene-related [4]. Recent studies have identified several hundred genes associated with epilepsy, among which genes encoding ion channels/receptors predominate [3,5]. As a Cl − channel and the primary mediator of inhibitory  17]. The receptor's agonist, GABA, binds to the agonist-binding site interfacing the extracellular N-terminal domains of the α and β subunits ( Figure 1B). The agonist binding triggers the conformational opening of the receptor channel and leads to chloride anion influx through the ion pore formed by the transmembrane segments of all five subunits [13,17,18]. Due to the pentameric composition of the GABAAR, mutations of the amino acid residues in many different GABAAR subunits have been reported to alter the function, expression level, and subcellular distributions of the receptor [19][20][21][22].  (A) Diagram represents the T292 residue located at GABRA1 TM2. (B) Cryo-EM structure of the GABA A receptor (PDB ID 6X3U) with highlighted T292 residue (yellow) and the bond agonist GABA (orange). As indicated, the T292 residue forms part of the inner channel pore. (C) Depiction of the two de novo missense variants in the patients and unaffected parents. (D) Comparison of the GABRA1 protein from several species indicates that Thr292 (in bold red) and nearby amino acids are evolutionarily conserved. Protein sequences were acquired from Uniprot and Ensembl.

Phenotypic Comparisons of the Two Patients Carrying the GABRA1 Mutants
Our patient was a 2-year-old boy with severe neurodevelopmental delays manifested as visual impairment, feeding difficulties, and significantly below-average body weight and organ mass. He had, however, no diagnosed overt seizure activity despite a few abnormal electroencephalography (EEG) discharges observed. In addition, there was no abnormality that could be found in MRI. No similar neurodevelopmental phenotype has been observed in his siblings or family members. Through whole-exome sequencing (WES), we identified a novel de novo T292S (C875G) missense variant of GABRA1. The clinical phenotypes, particularly the lack of any observable somatic seizure activity, are drastically different from that of a previously reported patient who carried a de novo variant at the same residue T292I (C875T) and had the main clinical features of Lennox-Gastaut syndrome with light-sensitive myoclonic epilepsy, generalized tonic-clonic seizures, and developmental delay [19,31]. These strikingly different clinical phenotypes may implicate distinct impacts on the receptor functions by these two mutations albeit occurring on the same residue.

Structural Analysis and the Location of T292 Residue in GABRA1
The α1. T292 residue is located in the second transmembrane domain (TM2) of the α 1 subunit ( Figure 1A) and lies within the inner channel pore ( Figure 1B). The parents of both patients have normal alleles, but the patients have the heterozygous missense variant of the same site in GABRA1 (NM_000806.5: C.875C>G or 875C>T, NP_000797.2: p.Thr292Ser or Thr292Ile; Figure 1C). Sequence alignment demonstrates that this residue is highly conserved among different species, including human, monkey, rat, mouse, bovine, chicken, xenopus, and zebrafish species, implying the importance of this residue ( Figure 1D). This is in good agreement with previous studies reporting that modification or pathogenic variant of the channel-lining residues could lead to dramatic alterations of the receptor function, particularly channel-gating properties [32,33].

The T292S and T292I Variants Bidirectionally Affect GABA-Evoked Responses
The drastic differences in the clinical phenotypes of the two mutations may imply significantly different impacts on the GABA A R function. Since the functional impacts of neither variant have previously been fully characterized, we employed whole-cell patchclamp recordings of GABA-evoked currents to test whether the variants would affect GABA A R functions. The evoked whole-cell currents were measured by fast perfusion (2 s) of GABA with concentrations ranging from 0.1 µM to 1 mM at a holding membrane potential of −60 mV in HEK293 cells transiently overexpressing wild-type α1/β2/γ2, α1 T292S /β2/γ2, or α1 T292I /β2/γ2.
For the wild-type GABA A Rs, GABA-evoked inward currents increased in a dosedependent manner, in which the currents were observable at 1 µM and reached the maximum level at around 100 µM. In comparison with the wild-type receptor, the GABA sensitivity of the α1 T292S GABA A Rs was significantly increased (Figure 2). The currents gated by the mutated receptor were observable at the concentration of 0.1 µM of GABA and reached the maximal at around 10 µM of GABA ( Figure 2A). When we normalized the currents to the cell's own maximum response, we found that in comparison with the wild-type GABA A Rs, T292S GABA A Rs had shifted its dose-response curve to the left ( Figure 2C) and lowered its EC50 by more than ten-fold (T292S: 0.2895 µM; WT: 3.658 µM) ( Figure 2C,E). However, when we normalized the currents of the T292S GABA A Rs (3287 ± 422.5 pA) against the maximum response of WT (3075 ± 526.0 pA), we observed that the maximum response of the T292S GABA A Rs was not affected ( Figure 2B,D). To mimic the patient's heterozygous expression of the mutant subunit, we co-transfected both the wild-type α1 T292 and the α1 T292S mutant in a 1:1 ratio along with the β2 and γ2 subunits in HEK293 cells. We found that this heterozygous expression caused a shift of the GABA-evoked dose-response curve toward the midpoint between the wild-type and the homozygous curves (Supplemental Figure S1).
In stark contrast, the expression of the T292I mutant resulted in significant impairment of GABA A R function. It severely reduced the peak current amplitude at each concentration tested ( Figure 2A) with a drastically decreased maximum response (1033 ± 286.5 pA, which was only about one-third of the response of the wild-type or T292S GABA A Rs) ( Figure 2B,D). The reduced maximum response was also associated with a rightwardshifted dose-response curve and a more than 18-fold increased EC50 over the wild-type GABA A R (T292I: 69.28 µM; WT: 3.658 µM) ( Figure 2C-E).
To check whether either of the two variants could affect the ion selectivity of GABA A R, chloride ion channel amplitudes of the GABA-evoked peak currents were measured at stepwise membrane potentials ranging from −80 to +80 mV with 20 mV intervals. As shown in Figure 2F,G, we found that neither variant compromised the ion selectivity since the reversal potentials of both variant receptors were similar to that of wild-type receptors, in which all were at approximately 0 mV.  and mutant (n = 5) GABAAR-gated currents evoked by 1mM GABA. Statistical differences were determined using Student's t-test (*** p < 0.001, ns = not significant).
In stark contrast, the expression of the T292I mutant resulted in significant impairment of GABAAR function. It severely reduced the peak current amplitude at each concentration tested ( Figure 2A) with a drastically decreased maximum response (1033 ± 286.5 pA, which was only about one-third of the response of the wild-type or T292S GABAARs) ( Figure 2B,D). The reduced maximum response was also associated with a rightward-shifted dose-response curve and a more than 18-fold increased EC50 over the wild-type GABAAR (T292I: 69.28 μM; WT: 3.658 μM) ( Figure 2C-E).
To check whether either of the two variants could affect the ion selectivity of GABAAR, chloride ion channel amplitudes of the GABA-evoked peak currents were measured at stepwise membrane potentials ranging from −80 to +80 mV with 20 mV intervals. GABA-evoked responses of WT (black), α1 T292S (red), and α1 T292I (blue) of GABA A R show bidirectional changes in two de novo mutations at the same T292 residue of GABRA1. (A) Representative traces of GABA-evoked currents of WT, α1 T292S , and α1 T292I of GABA A R. GABA application (2 s) is indicated as a bold black line above the traces. (B) Maximum response of α1 T292S (n = 10) and α1 T292I (n = 12) of GABA A R in comparison with WT (n = 8). (C) Dose-response curves of GABA-evoked responses of WT (n = 8), α1 T292S (n = 10), and α1 T292I (n = 12) of GABA A R. The peak current amplitude at each GABA concentration was normalized to the maximum response (1 mM) of each receptor, respectively. (D) Dose-response curves of GABA-evoked responses of WT (n = 8), α1 T292S (n = 10), and α1 T292I (n = 12) of GABA A R. The peak current amplitude at each GABA concentration was normalized to the maximum response (1 mM) of WT. (E) EC50 of GABA-evoked response of WT, α1 T292S , and α1 T292I GABA A R. (F) Representative trace of I-V curve. (G) I-V curves for WT (n = 5) and mutant (n = 5) GABA A R-gated currents evoked by 1mM GABA. Statistical differences were determined using Student's t-test (*** p < 0.001, ns = not significant).

Neither T292S nor T292I Variants Affected GABA A R Total/Surface Expressions
The marked functional alterations of the two variants in the GABA-evoked currents were quite intriguing and prompted us to probe the mechanisms underlying the bidirectional changes. Functional changes could be due to changes in receptor expression or channel-gating properties. Since alteration in the level of GABA A Rs expression in the cell and/or on the cell membrane, including other previously identified GABA A R variants [23,34,35], have previously been reported to be important methods of regulating the functions of the receptors with physiological and pathogenic consequences, we first examined the possibility that these two variants would exert their impacts on the receptor by affecting the receptor total and/or surface expression levels. We quantified the total receptors expressed in the cell by immunoblotting total cell lysates and quantified the surface receptors expressed on the plasma membrane by specifically immunoblotting biotinylated surface receptors ( Figure 3) in HEK cells transiently expressing wild-type α1/β2/γ2, α1 T292S /β2/γ2, or α1 T292I /β2/γ2. To our surprise, unlike other previously identified GABA A R variants [19,20,22], we found that in comparison with the WT receptors, neither variant produced any significant alteration in either the total or the surface expression levels of the receptor ( Figure 3A-D).
The marked functional alterations of the two variants in the GABA-evoked currents were quite intriguing and prompted us to probe the mechanisms underlying the bidirectional changes. Functional changes could be due to changes in receptor expression or channel-gating properties. Since alteration in the level of GABAARs expression in the cell and/or on the cell membrane, including other previously identified GABAAR variants [23,34,35], have previously been reported to be important methods of regulating the functions of the receptors with physiological and pathogenic consequences, we first examined the possibility that these two variants would exert their impacts on the receptor by affecting the receptor total and/or surface expression levels. We quantified the total receptors expressed in the cell by immunoblotting total cell lysates and quantified the surface receptors expressed on the plasma membrane by specifically immunoblotting biotinylated surface receptors (Figure 3) in HEK cells transiently expressing wild-type α1/β2/γ2, α1T292S/β2/γ2, or α1T292I/β2/γ2. To our surprise, unlike other previously identified GABAAR variants [19,20,22], we found that in comparison with the WT receptors, neither variant produced any significant alteration in either the total or the surface expression levels of the receptor ( Figure 3A-D).  . T292S and T292I variants did not affect GABA A R total/surface expressions. (A) Representative blots of surface biotinylation of WT, α1 T292S , and α1 T292I of GABA A R and non-transfected HEK293. (B) Quantification of the surface α1 subunit expression normalized to Na + /K + ATPase (n = 9). (C) Representative blots of the total expression levels from WT, α1 T292S , and α1 T292I of GABA A R and non-transfected HEK293 samples. (D) Quantification of the total α1 subunit expression normalized to heat shock protein 90 (HSP90) (n = 7). Statistical differences were determined using one-way ANOVA (ns = not significant).

T292S and T292I Variants Bidirectionally Altered GABA A R Single Channel Properties
Since neither T292S nor T292I mutation affected GABA A R expression in the cell or on the plasma membrane as shown above, we next examined the possibility that the functional alterations resulted from the changes in channel-gating properties. We employed cellattached, single-channel recordings of the WT and mutant GABA A Rs transiently expressed in HEK cells with subsaturating (3 µM) or saturating (1 mM) concentrations of GABA contained in the recording pipette. As shown in Figure 4A-E, we found that the T292S mutant significantly increased GABA-evoked single-channel activity at subsaturating, but not at sat-

T292S and T292I Variants Resulted in Different Changes of GABA A R-Gated Tonic and Leak Currents
We noticed in a recent study by Butler, K.M., et al. that a T292K variant of the α 2 subunit caused the tonic opening of GABA A Rs [22]. Since the α 1 and α 2 subunits are highly homologous, we wondered whether the T292S and T292I variants in the α1 subunit would cause any changes to the tonic/leak currents of GABA A Rs. The GABA tonic current is the current gated through the GABA A R that is activated by the ambient GABA in the extracellular solution, and it can be revealed by blocking the GABA A R with the competitive antagonist bicuculline [36]. The leak current is the current gated through the channel pore of the non-activated GABA A R and is usually revealed with a receptor channel pore blocker, such as picrotoxin [37]. We found that 20 µM of bicuculline was already enough to block all tonic current because the application of a higher concentration (50 µM) did not induce a further blockade (Supplemental Figure S2A,B). Bath application of the saturated bicuculline (20 µM) produced pronounced tonic currents as evident by the upward baseline current drifting in cells expressing T292S GABA A Rs (249 ± 52.6 pA). In great contrast, bicuculline application resulted in only minimal or no tonic currents in cells expressing either the WT (13.63 ± 5.641 pA) or T292I (0.3709 ± 0.8945 pA) receptors ( Figure 5A,B). Similarly, bath application of picrotoxin (100 µM) revealed a notable leak current as evident with a small additional increase upward current on top of the bicuculline-induced tonic currents only in cells expressing the T292S GABA A Rs (72.22 ± 16.44 pA), but not in cells expressing the WT (1.586 ± 2.496 pA) or T292I (1.913 ± 2.304 pA) GABA A Rs ( Figure 5A,C). These results demonstrated that the T292S variant could increase both tonic and leak currents of GABA A Rs while the T292I variant could not affect either current.

Thiocolchicoside Restores the Function of T292S GABA A R
The increased GABA sensitivity of T292S variant is a surprising result since it suggests that a gain-of-function mutant of the GABA A R is also pathogenic. We reasoned that negative GABA A R modulators may be able to normalize the function of the mutant receptor back to the WT receptor level, thereby exhibiting therapeutic potential to treat patients carrying this gain-of-function GABA A R variant. We examined the effect of three clinically approved negative allosteric GABA A R modulators, bemegride [38], flumazenil [39], and thiocochicoside (TCC) [30], on the whole-cell currents evoked with perfusion of a range of GABA from 0.1 µM to 1 mM. We found that among the three drugs tested at a concentration of 200 µM, only TCC was able to significantly reduce the increased GABA currents gated through the T292S variant ( Figure 6A-D). We then tested the dose-response curve of TCC in decreasing T292S GABA A R gated currents evoked by 1 µM GABA in an effort to find an optimal concentration of TCC that was capable of fully restoring the GABA-sensitivity of T292S GABA A R ( Figure 6A,E and Supplemental Figure S3A,B). As shown in Figure 6A,E, we found that 1 µM TCC was able to fully shift the dose-response curve leftward to the level of the WT GABA A R.

Combination of Verapamil and Diazepam Partially Rescues the Function of T292I GABA A R
Since the T292I variant's loss-of-function is primarily a result of reduced channelopening probability, we reasoned that its channel abnormality may be restored by some positive GABA A R allosteric modulators, such as diazepam and verapamil, that have previously been shown to restore the function of some pathogenic loss-of-function GABA A R variants via improving channel-gating properties [20]. We first pretreated HEK293 cells transiently expressing the α1 T292I /β2/γ2 receptors with verapamil (4 µM) for 24 h, as well as acutely applied verapamil during the recordings of the full dose response of the GABA-evoked currents. As shown in Figure 7A,C, verapamil showed slight rescuing effects of the T292I GABA A R. We repeated the experiment using diazepam [40] and found that it also only showed partial rescuing effects ( Figure 7A,B). Then, we considered the possibility that a combination of verapamil and diazepam would result in a synergistic rescuing effect. Indeed, acute application of diazepam and verapamil together after chronic treatment of verapamil showed the best effect in shifting the GABA-evoked dose-response curve toward that of the WT counterpart ( Figure

T292S and T292I Variants Resulted in Different Changes of GABAAR-Gated Tonic and Leak Currents
We noticed in a recent study by Butler, K.M., et al. that a T292K variant of the α2 subunit caused the tonic opening of GABAARs [22]. Since the α1 and α2 subunits are highly homologous, we wondered whether the T292S and T292I variants in the α1 subunit would cause any changes to the tonic/leak currents of GABAARs. The GABA tonic current is the current gated through the GABAAR that is activated by the ambient GABA in the extracellular solution, and it can be revealed by blocking the GABAAR with the competitive antagonist bicuculline [36]. The leak current is the current gated through the channel pore

Thiocolchicoside Restores the Function of T292S GABAAR
The increased GABA sensitivity of T292S variant is a surprising result since it suggests that a gain-of-function mutant of the GABAAR is also pathogenic. We reasoned that negative GABAAR modulators may be able to normalize the function of the mutant receptor back to the WT receptor level, thereby exhibiting therapeutic potential to treat patients carrying this gain-of-function GABAAR variant. We examined the effect of three clinically approved negative allosteric GABAAR modulators, bemegride [38], flumazenil [39], and thiocochicoside (TCC) [30], on the whole-cell currents evoked with perfusion of a range of GABA from 0.1 μM to 1 mM. We found that among the three drugs tested at a concentration of 200 μM, only TCC was able to significantly reduce the increased GABA currents gated through the T292S variant ( Figure 6A-D). We then tested the dose-response curve of TCC in decreasing T292S GABAAR gated currents evoked by 1 μM GABA in an effort to find an optimal concentration of TCC that was capable of fully restoring the GABAsensitivity of T292S GABAAR ( Figure 6A,E and Supplemental Figure S3A,B). As shown in Figure 6A,E, we found that 1 μM TCC was able to fully shift the dose-response curve leftward to the level of the WT GABAAR.

Combination of Verapamil and Diazepam Partially Rescues the Function of T292I GABAAR
Since the T292I variant's loss-of-function is primarily a result of reduced channelopening probability, we reasoned that its channel abnormality may be restored by some positive GABAAR allosteric modulators, such as diazepam and verapamil, that have previously been shown to restore the function of some pathogenic loss-of-function GABAAR variants via improving channel-gating properties [20]. We first pretreated HEK293 cells transiently expressing the α1T292I/β2/γ2 receptors with verapamil (4 μM) for 24 h, as well as acutely applied verapamil during the recordings of the full dose response of the GABAevoked currents. As shown in Figure 7A,C, verapamil showed slight rescuing effects of the T292I GABAAR. We repeated the experiment using diazepam [40] and found that it also only showed partial rescuing effects ( Figure 7A,B). Then, we considered the possibility that a combination of verapamil and diazepam would result in a synergistic rescuing effect. Indeed, acute application of diazepam and verapamil together after chronic treatment of verapamil showed the best effect in shifting the GABA-evoked dose-response curve toward that of the WT counterpart ( Figure 7A,D), although the shift remained incomplete.

Discussion
In this study, we reported a novel de novo missense variant T292S of the T292 residue of GABRA1 and compared it with a previously found de novo missense variant T292I of the same residue. As summarized in Table 1, clinical observations and laboratory investigations revealed drastically distinct clinical phenotypes and functional alterations in these GABA A Rs mutants. The patient carrying the T292S variant is featured with developmental delay without observable somatic seizure activity until deceased (while the possibility that epileptic activity could develop at a later developmental stage cannot be entirely ruled out); while the patient with the T292I variant primarily manifested with severe epilepsy [19,31]. Considering the fact the T292 residue is highly conserved among species and is one of the residues lining the channel pore, functional characterization of these variants identified from patients may improve our understanding of the underlying pathophysiology. Our functional analysis performed in HEK293 cells showed that the T292S and T292I variants of the GABA A R α1 subunit conferred opposing changes in GABA agonist sensitivity and potency. The T292S variant induced a leftward shift of the GABA dose-response curve and lowered the EC 50 of GABA without altering the maximum response. On the other hand, the T292I variant caused a rightward shift of the GABA dose-response curve, increased the EC 50 of GABA, and reduced the maximum response. To our knowledge, this is the first study that performs detailed functional characterizations of the T292S and T292I variants, and importantly, the first to report that the same subunit residue of GABA A R replaced by different amino acids would result in opposite impacts on the receptor functions. that epileptic activity could develop at a later developmental stage cannot be entirel ruled out); while the patient with the T292I variant primarily manifested with severe epi lepsy [19,31]. Considering the fact the T292 residue is highly conserved among specie and is one of the residues lining the channel pore, functional characterization of thes variants identified from patients may improve our understanding of the underlying path ophysiology. Our functional analysis performed in HEK293 cells showed that the T292 and T292I variants of the GABAAR α1 subunit conferred opposing changes in GABA ag onist sensitivity and potency. The T292S variant induced a leftward shift of the GABA dose-response curve and lowered the EC50 of GABA without altering the maximum re sponse. On the other hand, the T292I variant caused a rightward shift of the GABA dose response curve, increased the EC50 of GABA, and reduced the maximum response. To ou knowledge, this is the first study that performs detailed functional characterizations of th T292S and T292I variants, and importantly, the first to report that the same subunit resi due of GABAAR replaced by different amino acids would result in opposite impacts o the receptor functions. that epileptic activity could develop at a later developmental stage cannot be entirel ruled out); while the patient with the T292I variant primarily manifested with severe ep lepsy [19,31]. Considering the fact the T292 residue is highly conserved among specie and is one of the residues lining the channel pore, functional characterization of thes variants identified from patients may improve our understanding of the underlying path ophysiology. Our functional analysis performed in HEK293 cells showed that the T292 and T292I variants of the GABAAR α1 subunit conferred opposing changes in GABA ag onist sensitivity and potency. The T292S variant induced a leftward shift of the GABA dose-response curve and lowered the EC50 of GABA without altering the maximum re sponse. On the other hand, the T292I variant caused a rightward shift of the GABA dose response curve, increased the EC50 of GABA, and reduced the maximum response. To ou knowledge, this is the first study that performs detailed functional characterizations of th T292S and T292I variants, and importantly, the first to report that the same subunit res due of GABAAR replaced by different amino acids would result in opposite impacts o the receptor functions. Channel open probability the same residue. As summarized in Table 1, clinical observations and laboratory inv gations revealed drastically distinct clinical phenotypes and functional alterations in th GABAARs mutants. The patient carrying the T292S variant is featured with developme delay without observable somatic seizure activity until deceased (while the possib that epileptic activity could develop at a later developmental stage cannot be enti ruled out); while the patient with the T292I variant primarily manifested with severe lepsy [19,31]. Considering the fact the T292 residue is highly conserved among spe and is one of the residues lining the channel pore, functional characterization of th variants identified from patients may improve our understanding of the underlying p ophysiology. Our functional analysis performed in HEK293 cells showed that the T2 and T292I variants of the GABAAR α1 subunit conferred opposing changes in GABA onist sensitivity and potency. The T292S variant induced a leftward shift of the GA dose-response curve and lowered the EC50 of GABA without altering the maximum sponse. On the other hand, the T292I variant caused a rightward shift of the GABA do response curve, increased the EC50 of GABA, and reduced the maximum response. To knowledge, this is the first study that performs detailed functional characterizations o T292S and T292I variants, and importantly, the first to report that the same subunit due of GABAAR replaced by different amino acids would result in opposite impact the receptor functions. of GABRA1 and compared it with a previously found de novo missense variant T292I the same residue. As summarized in Table 1, clinical observations and laboratory inve gations revealed drastically distinct clinical phenotypes and functional alterations in the GABAARs mutants. The patient carrying the T292S variant is featured with developmen delay without observable somatic seizure activity until deceased (while the possibil that epileptic activity could develop at a later developmental stage cannot be entir ruled out); while the patient with the T292I variant primarily manifested with severe e lepsy [19,31]. Considering the fact the T292 residue is highly conserved among spec and is one of the residues lining the channel pore, functional characterization of the variants identified from patients may improve our understanding of the underlying pa ophysiology. Our functional analysis performed in HEK293 cells showed that the T29 and T292I variants of the GABAAR α1 subunit conferred opposing changes in GABA a onist sensitivity and potency. The T292S variant induced a leftward shift of the GAB dose-response curve and lowered the EC50 of GABA without altering the maximum sponse. On the other hand, the T292I variant caused a rightward shift of the GABA dos response curve, increased the EC50 of GABA, and reduced the maximum response. To o knowledge, this is the first study that performs detailed functional characterizations of t T292S and T292I variants, and importantly, the first to report that the same subunit re due of GABAAR replaced by different amino acids would result in opposite impacts the receptor functions. and is one of the residues lining the channel por variants identified from patients may improve our ophysiology. Our functional analysis performed in and T292I variants of the GABAAR α1 subunit con onist sensitivity and potency. The T292S variant i dose-response curve and lowered the EC50 of GA sponse. On the other hand, the T292I variant cause response curve, increased the EC50 of GABA, and re knowledge, this is the first study that performs deta T292S and T292I variants, and importantly, the fir due of GABAAR replaced by different amino acid the receptor functions.  [19,31]. Considering the fact the T292 resid and is one of the residues lining the channel por variants identified from patients may improve our ophysiology. Our functional analysis performed i and T292I variants of the GABAAR α1 subunit con onist sensitivity and potency. The T292S variant dose-response curve and lowered the EC50 of GA sponse. On the other hand, the T292I variant cause response curve, increased the EC50 of GABA, and r knowledge, this is the first study that performs det T292S and T292I variants, and importantly, the fir due of GABAAR replaced by different amino acid the receptor functions.

GABAARs
mutants. The patient carrying the T292S variant is fea delay without observable somatic seizure activity until decea that epileptic activity could develop at a later developmenta ruled out); while the patient with the T292I variant primarily m lepsy [19,31]. Considering the fact the T292 residue is highly and is one of the residues lining the channel pore, functiona variants identified from patients may improve our understandi ophysiology. Our functional analysis performed in HEK293 ce and T292I variants of the GABAAR α1 subunit conferred oppo onist sensitivity and potency. The T292S variant induced a le dose-response curve and lowered the EC50 of GABA without sponse. On the other hand, the T292I variant caused a rightwar response curve, increased the EC50 of GABA, and reduced the m knowledge, this is the first study that performs detailed function T292S and T292I variants, and importantly, the first to report t due of GABAAR replaced by different amino acids would resu the receptor functions. gations revealed drastically distinct clinical phen GABAARs mutants. The patient carrying the T29 delay without observable somatic seizure activ that epileptic activity could develop at a later ruled out); while the patient with the T292I vari lepsy [19,31]. Considering the fact the T292 re and is one of the residues lining the channel p variants identified from patients may improve o ophysiology. Our functional analysis performe and T292I variants of the GABAAR α1 subunit c onist sensitivity and potency. The T292S varia dose-response curve and lowered the EC50 of G sponse. On the other hand, the T292I variant cau response curve, increased the EC50 of GABA, an knowledge, this is the first study that performs d T292S and T292I variants, and importantly, the due of GABAAR replaced by different amino a the receptor functions. There are several potential reasons that can explain these functional changes, which were examined in detail in our experiments [22]. First, the genetic mutations observed in the GABARA1 variants may theoretically alter the receptor expression, assembly, and trafficking that lead to the change of the surface/total receptor numbers [20]. However, we found that neither T292S nor T292I variants caused changes in the receptor numbers as evidenced by the results of immunoblotting and surface biotinylation assays.
Second, single amino acid mutations may induce structural changes in the GABA agonist-binding pocket and result in increased or decreased GABA binding affinity [22,24,41]. However, as the T292 residue is located at the inner channel pore, which is located far from the GABA binding region, the chance that the mutation causes an allosteric change in the structure of the GABA agonist-binding pocket is low.
Third, since this residue lies in the middle of the channel pore, its mutations may produce conformational changes that lead to altered GABA sensitivity in inducing channel gating [24,28]. Supporting this conjecture, our single-channel recording data showed significant changes in channel-gating properties in both variants. The T292S variant increased the single-channel open time and open probability under subsaturating (3 µM) and saturating (1 mM) GABA concentrations, suggesting the alteration of GABA A R function by increasing the sensitivity of GABA to induce channel opening. In contrast, the T292I variant showed a significant decrease in GABA's ability to keep the channel opened as evident by the lack of opening activity at the subsaturating GABA level, and significantly reduced channel open time and open probability at the saturating GABA stimulation (1mM). Together, these results strongly suggest that residue T292 in the α1 subunit has a critical role in determining the channel open threshold by its agonists. The idea that the opposing functional alterations of these two mutations primarily resulted in changing GABA's sensitivity in gating the receptor channel is further supported by the results of our tonic and leak currents assay. We found that the T292S (but not T92I) GABA A R showed increased tonic currents and leak currents. We postulated that the T292S variant not only decreases the threshold of GABA-induced channel opening, but also keeps the channel at a partially opened status even in the absence of agonist binding, whereas the T292I variant turns the channel into a more closed status by decreasing the channel opening ability. Given that the T292 residue is located at the TM2 segment and forms part of the channel pore, it should not be surprising to see such changes in the channel-gating properties. In particular, the T292I variant involves a threonine to isoleucine mutation, which may result in a significant change in the side chain's length and polarity that are not in favour of channel opening. Similarly, as the T292S variant involves a more subtle threonine to serine mutation, it results in less alteration in the side chain's length and polarity, and such relative subtle changes may, in turn, favour channel opening. Together, the ability of both T292S and T292I to dramatically and oppositely change the channel open probability demonstrates that α1T292 is a critical residue for controlling GABA A R channel gating.
Since we identified the first GABRA1 mutation, A322D, in an epilepsy patient two decades ago [42], there have been more than 25 GABRA1 mutations that are related to pediatric encephalopathy reported, such as T20I, V74I, S76R, F104C, R112Q, N115D, L146M, T156C, P181S, D219N, P260L, S270H, V287L, T292I, K306T A322D, A332V, as well as R214H/C, G251S/D, M263T/I, and T289P/A, in which the same residue is replaced by different amino acids [8,13,21,25,35,43]. Most of these characterized mutations cause a loss of function of the GABA A R due to reduced surface/total expression, decreased agonist sensitivity, or compromised channel gating [20,42,44,45]. In addition to our T292S variant reported here, only one of these GABRA1 variants, A332V, which is located in the channel pore-forming TM3, is recently reported to enhance the receptor function [41]. In the present work, we report for the first time that the T292S variant, unlike most of the previously reported pathogenic GABA A R variants, is clearly a gain-of-function variant; while the T292I variant, which occurs at the same residue with a different amino acid substitution as T292S, is a loss-of-function variant.
Following functional profiling of the pathological mechanisms of the T292S and T292I mutations, we also attempted to identify potential therapeutic options for more effective and personalized treatments. Guided by the detailed functional phenotypes of the two variants, we performed a quick screening of some clinically approved drugs that directly or indirectly act on GABA A Rs that may potentially reduce the functional abnormalities of certain mutants of GABA A Rs. Thiocochicoside (TCC) is used in the clinic as a muscle relaxant, but it is also an allosteric GABA A R inhibitor that shows potent antagonistic effects against GABA A R [30]. At a 1 µM concentration, TCC could fully restore the increased GABA-evoked dose response of the T292S variant to the WT level, with superior effects over other GABA A R negative allosteric modulators, bemegride and flumazenil. Our study, along with other previous studies, suggest that TCC can be a potential therapeutic option for gain-of-function GABA A R variants [22,41].
In contrast, we found that the loss-of-function variant T292I GABA A R-gated currents could be partially restored by a combination of two positive GABA A R modulating drugs, diazepam and verapamil. Chronic treatment of verapamil was shown to enhance channel gating with an elongated open time and increased open probability of R214C GABA A R, suggesting diazepam and verapamil may work synergistically to ultimately improve the function of the T292I variant GABA A Rs. Our study indicates that these mutations have opposite impacts on the function of GABA A Rs, and although occurring at the same residue, thereby require different functional and pharmacological strategies to restore the function of the receptor to the level of their WT counterpart. Thus, our findings suggest that it is of paramount importance to perform functional and pharmacological analysis after exome sequencing to determine the pathological mechanisms. This would aid in searching for the appropriate therapeutic options for patients carrying a de novo mutation of principle neurotransmitter receptors, such as the GABA A Rs.
In conclusion, our functional analyses of the two de novo variants, T292S and T292I, of the same GABRA1 T292 residue from two patients with distinct clinical phenotypes have revealed their gain-of-function and loss-of-function impacts on the GABA A R, respectively. The study not only provides evidence for the pathogenic contributions of the two variants to the patients' pathology, but also indicates the crucial role of the T292 residue in controlling channel gating. There are around 30% of children with refractory epilepsy that do not respond to conventional drug treatments due to unknown functional alterations or unidentified causes [46,47]. Moreover, traditional anti-seizure drugs show little to no effect toward rescuing deficits caused by specific mutations in the GABA A R subunits [48,49], and in certain cases, can even exacerbate the symptoms [50,51]. The pharmacological characterizations in our study provide differential therapeutic suggestions for managing these two patients and endorse the importance of precision medicine for pediatric channelopathies.

Structural Modeling of GABA A Receptor
The crystal structure of the GABA A R was obtained from the Protein Data Bank (PDB) (ID: 6X3U) [52]. Molecular graphics and analyses were performed with the Py-MOL Academic-Professional version (Schrödinger, Inc., New York, NY, USA).

Plasmid Construction
The cDNAs encoding rat GABA A R α1, GABA A R β2, GABA A R γ2, and EGFP were separately cloned into the pcDNA3.0 vector (Invitrogen, Waltham, MA, USA). Using the plasmid with the WT rat GABA A R α1 cDNA as the template, PCR mediated QuickChange site-directed mutagenesis was performed with a high-fidelity Hot Start DNA polymerase (KAPA Biosystmems, Wilmington, MA, USA Cat: KM2605) to construct the mutated variants of the GABA A R α1 (c.875 C>G and c.875 C>T) subunit (point mutation primers: Sense 5 -cgaccgttctgagcatgacaacctt-3 , reverse 5 -aaggttgtcatgctcagaacggtcg-3 and sense 5 -cgaccgttctgatcatgacaacctt-3 , reverse 5 -aaggttgtcatgatcagaacggtcg-3 for T292S and T292I mutations, respectively). The mutations were confirmed by sequencing, and the resulting plasmids encoding the mutant GABA A R α1 were referred to as α1 T292S (T292S) and α1 T292I (T292I).

Western Blot and Surface Biotinylation
Transfected HEK293 cells in a 6-well cell culture plate were washed with ice-cold PBS three times and lysed with 0.5 mL of 1% SDS TBS buffer containing a protease inhibitor cocktail (Bimake, Houston, TX, USA) at 4 • C for 30 min. Cells were then harvested and homogenized using needles with gauge sizes from 18 G to 23 G to 26 G, progressively. The supernatant was collected after centrifugation (13,000 rpm, 4 • C, 10 min) and the protein concentration in each sample was measured using the Pierce BCA protein assay (Thermo Scientific, Waltham, MA, USA, REP233228). Samples containing equal amounts of the total protein were treated with the 6X Laemmli sample buffer containing 9% betamercaptoethanol (MilliporeSigma, St. Louis, MO, USA) and heated at 55 • C for 5 min before loading onto 10% SDS-PAGE gels. The proteins in the gels were transferred onto a PVDF membrane (MilliporeSigma, St. Louis, MO, USA). Rabbit anti-GABA A Rα1 polyclonal antibody (1:1000, MilliporeSigma, St. Louis, MO, USA, Cat. #06-868) was used to detect the WT and variant GABA A R α1 subunits. HSP90 (mouse monoclonal antibody 1:4000, BD, Biosciences, Franklin Lakes, NJ, USA, Cat. #610418) served as the loading control. The HRP-conjugated anti-mouse or anti-rabbit secondary antibodies (Thermo-Fisher Scientific, Waltham, MA, USA, Cat. # 31430 and 31460) were used in a ratio of 1:5000. The signal was detected using an ECL detection system (MilliporeSigma, St. Louis, MO, USA, Immobilon Crescendo Western HPP Substrate, Cat#WBLUR0500) via Bio-Rad ChemiDoc MP imaging system.
In the surface biotinylation assay, cells in a 6-well cell culture plate were washed 3 times on ice with ice-cold PBS 40 h post-transfection, then 0.5 mL of PBS with the membrane-impermeable reagent EZ-Link Sulfo-HNS-LC-Biotin (1 mg/mL, Thermo Scientific, Waltham, MA, USA) was added into each well and kept at 4 • C for 30 min to label surface membrane proteins. To quench the biotin reaction, cells were washed on ice with 1 mL of 100 mM glycine dissolved in ice-cold PBS three times (10 min each time) at 4 • C. Then the biotinylated cells in each well were harvested using 0.5 mL of 1% SDS lysate buffer (1% SDS in 1XTris Buffered saline, pH 7.6)) at 4 • C for 30 min and homogenized using needles with a gauge size from G18 to G23 to G26, progressively. Supernatants were collected after centrifugation (13,000 rpm, 4 • C, 10 min). The protein concentration in each sample was measured using the Pierce BCA protein assay (Thermo scientific, Waltham, MA, USA REP233228). Each sample with an equal amount of biotin-labeled membrane proteins was rotationally incubated with 30 uL of High-Capacity Streptavidin Agarose Resin (beads) (Thermo Scientific, Waltham, MA, USA REF20359) at 4 • C overnight, after which the beads were pulled down and washed with PBS for 3-4 times via centrifugation (500 g, 4 • C, 2 min). The beads were suspended with 45 uL of 2X Laemmli sample buffer and heated at 55 • C for 5 min. Heated samples were then centrifuged at 1000 rpm for 2 min. Finally, the supernatants were loaded onto 10% SDS-PAGE gels. Na + /K + ATPase (abcam: Mouse monoclonal antibody alpha1 sodium potassium ATPase ([464.6] 1:1000) served as a loading control for the biotinylated membrane proteins and β-actin (MilliporeSigma, St. Louis, MO, USA: Mouse monoclonal antibody 1:3000, Cat, #A2228) served as a labeling control for the cytoplasmic proteins.

Electrophysiology
Electrophysiology experiments, including whole-cell voltage-clamp recordings and cell-attached single-channel recordings, were performed at room temperature on HEK293 cells transfected with the WT, T292S, or T292I GABA A R subunits. The recordings were conducted and low-pass-filtered at 2 kHz using a MultiClamp 700A amplifier (Molecular Devices, San Jose, CA, USA) and digitized at 20 kHz using Digidata 1440A. The recordings were performed using Clampex 10.7 software (Molecular Devices, San Jose, CA, USA) and the data were analyzed offline using Clampfit 10.7 (Molecular Devices, San Jose, CA, USA).
Tonic and leak currents assay was performed using a bath solution exchange protocol. Specifically, cells were patched and then recorded using a gap-free program, where 10 µM GABA, GABA-free ECS, 20 µM bicuculline, and 20 µM bicuculline plus 100 µM picrotoxin were bath applied sequentially and the application of each solution lasted at least 60 s to allow enough time for solution exchanges in the chamber. The initial application of GABA was meant to confirm the success of transfection of the recombinant GABA A Rs.
Cell-attached single-channel recordings were conducted using the same protocol as before [15]. The patch electrodes were fire-polished to a resistance of 10-20 MΩ and filled with the extracellular recording solution with 3 µM or 1 mM GABA (pH = 7.4, 310-320 mOsm), and the holding potential was +100 mV. Single-channel events were detected using the 50% amplitude threshold detection method and were visually inspected before being accepted.

Chemicals
Picrotoxin (Abcam, Cambridge, UK), bemegride (Cedarlane, Burlington, ON, Canada), diazapem (Tocris Bioscience, Bristol, UK), and thiocolchicoside (Cedarlane, Burlington, ON, Canada) were first dissolved in DMSO for the stock solutions (100 mM) and further diluted in ECS to the desired working concentration during the experiment. Bicuculline (Hellobio Inc, Princeton, NJ, USA) and flumazenil (Tocris Bioscience, Bristol, UK) were first dissolved in DMSO for the stock solutions (20 mM) and further diluted in ECS during the experiment. Verapamil hydrochloride (Tocris Bioscience, Bristol, UK) was diluted in water to make a 4 mM stock solution.

Data Analysis
Data were presented as mean ± SEM (n = number of cells). The unpaired t-test, one-way ANOVA, and two-way ANOVA (both ANOVAs were followed by Bonferroni post hoc tests) were used for statistical analysis, and * p < 0.05, ** p < 0.01, and *** p < 0.001 were considered statistically significant. Dose-response curves were fitted by the Hill equation, and EC 50 was calculated by GraphPad prism 6 (GraphPad Software, San Diego, CA, USA). All data points were normalized to the maximum response of GABA-evoked current for each cell unless specified in the context. Whole-cell peak currents, the area under the curve, channel kinetics, and single-channel currents were analyzed by Clampfit 10.7(Molecular Devices, San Jose, CA, USA).