The synthetic cannabinoid dehydroxylcannabidiol restores the function of a major GABAA receptor isoform in a cell model of hyperekplexia

The functions of the glycine receptor (GlyR) and GABAA receptor (GABAAR) are both impaired in hyperekplexia, a neurological disorder usually caused by GlyR mutations. Although emerging evidence indicates that cannabinoids can directly restore normal GlyR function, whether they affect GABAAR in hyperekplexia remains unknown. Here we show that dehydroxylcannabidiol (DH-CBD), a synthetic nonpsychoactive cannabinoid, restores the GABA- and glycine-activated currents (IGABA and IGly, respectively) in HEK293 cells coexpressing a major GABAAR isoform (α1β2γ2) and GlyRα1 carrying a human hyperekplexia-associated mutation (GlyRα1R271Q). Using coimmunoprecipitation and FRET assays, we found that DH-CBD disrupts the protein interaction between GABAAR and GlyRα1R271Q. Furthermore, a point mutation of GlyRα1, changing Ser-296 to Ala-296, which is critical for cannabinoid binding on GlyR, significantly blocked DH-CBD-induced restoration of IGABA and IGly currents. This S296A substitution also considerably attenuated DH-CBD-induced disruption of the interaction between GlyRα1R271Q and GABAAR. These findings suggest that, because it restores the functions of both GlyRα1 and GABAAR, DH-CBD may represent a potentially valuable candidate drug to manage hyperekplexia.

GABA A R, another widely distributed inhibitory ligandgated ion channel in the central nervous system (26,27), has widely been proved to be colocalized with GlyR in spinal cord and brain stem neurons (28 -33). For instance, GlyR is colocalized with ␣ 1 and ␥ 2 subunits of GABA A R in the hypoglossal nucleus of mice (32). Double-immunofluorescence staining also showed that GABA A receptor-positive cells exhibit prominent glycine receptor immunoreactivity in spinal cord neurons (33). Emerging evidence indicates that GABA A R is functionally impaired in hyperekplexia disease (34,35). For instance, electrically evoked GABAergic inhibitory postsynaptic currents are decreased in the spinal cord of GlyR␣ 1 R271Q mutant mice (34). Consistent with this, we recently verified decreased activity of GABA A R in hyperekplexic transgenic mice carrying the R271Q or S267Q mutation (36). Such functional impairment can ultimately be attributed to the direct protein interaction between GABA A R and mutant GlyR␣ 1 . Benzodiazepines, first-line medications for treating hyperekplexia disease in the clinic (37)(38)(39), restored the function of GABA A R but not GlyR (40,41). However, medicines that target both receptors have not yet been developed.
As positive GlyR allosteric modulators, cannabinoids can directly potentiate GlyR function in a cannabinoid receptor 1 or 2 (CB1/2R)-independent manner (42)(43)(44)(45)(46)(47). Dehydroxylcannabidiol (DH-CBD), a synthetic nonpsychoactive cannabinoid, significantly alleviates the abnormal startle reflex and muscle stiffness in hyperekplexic transgenic mutant mice (25). Such effects were blocked by a point mutation (Ser3Ala) of GlyR␣ 1 Ser-296, which is essential for DH-CBD binding. Considering the restorative effects of cannabinoids on GlyR␣ 1 , we wondered whether it can restore the function of GABA A R in hyperekplexia disease. In this study, we coexpressed GABA A R (␣ 1 ␤ 2 ␥ 2 ) and R271Q mutant GlyR␣ 1 in a HEK293 cell model and combined various approaches, such as electrophysiological recording, FRET assays, and coimmunoprecipitation, to explore possible effects of DH-CBD on the function of GlyR␣ 1 and GABA A R in hyperekplexia disease.

DH-CBD restores the function of hyperekplexic mutant GlyR␣ 1
We first examined the effect of the R271Q mutation on GlyR␣ 1 function using single-cell patch clamp recording (Fig.  1A). The R271Q mutation significantly reduced the I Gly in HEK293 cells expressing GlyR␣ 1 R271Q in the presence and absence of GABA A R (Fig. 1B). DH-CBD is a synthetic nonpsychoactive cannabinoid modified from ⌬ 9 -tetrahydrocannabinol (48), a major component of cannabis (Fig. 1C). We then evaluated the effect of DH-CBD on I Gly in HEK293 cells coexpressing GlyR␣ 1 R271Q and GABA A R. Consistent with a previous report (25), DH-CBD at 1, 3, and 10 M dose-dependently caused a 10-to 30-fold increase in I Gly (Fig. 1D). Furthermore, DH-CBD also considerably restored the dose-response curve (Fig.  1E) and decreased the EC 50 values of GlyR␣ 1 R271Q (Fig. 1F).

R271Q
Compared with WT GlyR␣ 1 , coexpression of GlyR␣ 1 R271Q significantly decreased the GABA-activated current (I GABA ) in HEK293 cells ( Fig. 2A). DH-CBD at 3 M and 10 M, but not 1 M, remarkably restored I GABA (Fig. 2B). Additionally, GlyR␣ 1 R271Q obviously shifted the dose-response curve of I GABA to the right and increased the EC 50 values of GABA A R (Fig. 2, C and D). Such effects were eliminated by preincubation of DH-CBD (Fig. 2, C and D). DH-CBD-induced restoring effect on GABA A R seem to depend on GlyR␣ 1 because DH-CBD at 10 M could not affect I GABA in HEK293 cells expressing GABA A R alone (Fig. 2E). Previous reports showed that cannabidiol (CBD), another nonpsychoactive cannabinoid, could also potentiate GlyR function (46,47). We then incubated HEK293 cells with CBD. CBD at 10 M significantly restored GlyR␣ 1 R271Q -decreased I GABA in HEK293 cells (Fig. S1).

R271Q
We have reported previously that the decreased activity of GABA A R in hyperekplexia was due to the protein interaction between GABA A R and mutant GlyR␣ 1 (36). Such effects may specifically depend on the type of GlyR␣ 1 mutations because GlyR␣ 1 carrying a nonhyperekplexic K385A mutation showed very weak binding with GABA A R and had no effect on I GABA (Fig. S2), which was similar to WT GlyR␣ 1 . Next we performed a coimmunoprecipitation assay to measure the effects of DH-CBD on the protein interaction between both receptors using HEK293 cells coexpressing GlyR␣ 1 R271Q and GABA A R. Preincubation of DH-CBD significantly reduced the amount of GlyR␣ 1 R271Q protein coimmunoprecipitated with GABA A R (Fig. 3A).
Next we verified this effect of DH-CBD in HEK293 cells by FRET, a powerful technique for studying protein interactions in living cells with advanced spatial and temporal resolution (49). To perform the FRET assay, cyan fluorescent values of I Gly activated by 1 mM glycine in HEK293 cells expressing GABA A R (␣ 1 ␤ 2 ␥ 2 ) and/or R271Q mutant ␣ 1 GlyR (n ϭ 9). C, chemical structure of ⌬ 9 -tetrahydrocannabinol (THC) and DH-CBD. D, representative trace records and average values of I Gly activated by 1 mM glycine in HEK293 cells coexpressing GABA A R (␣ 1 ␤ 2 ␥ 2 ) and R271Q mutant ␣ 1 GlyR with or without 1 M, 3 M, and 10 M DH-CBD preincubation (n ϭ 7). E, dose-response curves of I Gly in HEK293 cells coexpressing GABA A R (␣ 1 ␤ 2 ␥ 2 ) and R271Q mutant ␣ 1 GlyR with or without 10 M DH-CBD preincubation. The data were normalized to I max of the WT GlyR group (n ϭ 6). F, EC 50 values of I Gly induced by increasing glycine concentrations in HEK293 cells coexpressing GABA A R (␣ 1 ␤ 2 ␥ 2 ) and R271Q mutant ␣ 1 GlyR with or without 10 M DH-CBD preincubation (n ϭ 6). Data are represented as mean Ϯ S.D. **, p Ͻ 0.01; ***, p Ͻ 0.001; based on unpaired t tests; ns, not significant (p Ͼ 0.05).

Cannabinoids restore GABA A R function in hyperekplexia
protein (CFP, the donor) and yellow fluorescent protein (YFP, the acceptor) were conjugated to the N-termini of the GlyR␣ 1 R271Q and GABA A R␣ 1 subunits separately (Fig. 3B). CFP-tagged GlyR␣ 1 R271Q and YFP-tagged GABA A R were functionally expressed in HEK293 cells (Fig. 3, C and D) and could be detected by a dual-channel spinning disk confocal microscope (Fig. S3). DH-CBD did not change the YFP/CFP emission ratio in HEK293 cells coexpressing CFP-tagged GlyR␣ 1 WT and YFP-tagged GABA A R (Fig. 3F). However, after 30 -60 min of incubation, DH-CBD significantly increased the YFP/CFP emission ratio in HEK293 cells coexpressing CFP-tagged GlyR␣ 1 R271Q and YFP-tagged GABA A R ( Fig. 3E), suggesting a changed protein interaction pattern between GlyR␣ 1 R271Q and GABA A R induced by DH-CBD treatment.

The S296A mutation diminishes DH-CBD-induced restoration of GlyR␣ 1 and GABA A R function in hyperekplexia
Ser-296 in the third transmembrane domain of the GlyR␣ 1 subunit has been suggested to be essential for DH-CBD-GlyR␣ 1 interaction (46,48,50). The S296A mutation in GlyR␣ 1 blocked the rescuing effects of DH-CBD on both I Gly and I GABA in HEK293 cells coexpressing GABA A R and GlyR␣ 1 carrying the R271Q and S296A mutations (Fig. 4, A and B). The S296A mutation also significantly inhibited the restoring effects of DH-CBD on the protein interaction between GlyR␣ 1 R271Q and GABA A R (Fig. 4C). Benzodiazepines are routine drugs for treating hyperekplexia in the clinic, targeting GABA A R (37,38,(51)(52)(53). However, unlike DH-CBD, diazepam could not affect the protein interaction between GABA A R and GlyR␣ 1 R271Q (Fig. 4D).

Discussion
Benzodiazepines are the most commonly used drugs for the treatment of hyperekplexia disease in the clinic (37, 38, 51-53).
Our recent study also reported that diazepam can restore the function of pre-and extrasynaptic GABA A R in hyperekplexia disease (36). Benzodiazepines are effective and relatively safe to treat this disease, especially at low doses (37)(38)(39). However, benzodiazepines may cause sedative effects even at low doses (54), which is far from ideal for long-term treatment and children. In addition, there is a chance of tolerance after months of benzodiazepine use, even at low doses (55). Therefore, there is a need to develop new therapeutic avenues targeting both GlyR␣ 1 and GABA A R with minimal side effects for the treatment of hyperekplexia disease. The data presented in this study provide evidence that DH-CBD may be a more appropriate candidate medicine for treating hyperekplexia disease under certain conditions, such as the GlyR␣ 1 R271Q mutation. DH-CBD restores the function of hyperekplexic mutant GlyR␣ 1 and GABA A R by interrupting the protein interaction between these two receptors, whereas benzodiazepines only restore GABA A R function (53).
Emerging evidence shows that Ser-296 of the GlyR ␣ 1 subunit is a critical site for cannabinoid action (46,48,50). In this study, the S296A site mutation blocked DH-CBD-induced disruption of the protein interaction between GlyR␣ 1 R271Q and GABA A R. Such an effect likely leads to restoration of GABA A R function in HEK293 cells. Considering the fact that DH-CBD had no effect on GABA A R alone expressed in HEK293 cells, we suppose that DH-CBD-induced functional restoration of GABA A R must be achieved by first acting on GlyR␣ 1 . However, the detailed mechanism of how DH-CBD diminishes the hijacking effect of mutant GlyR␣ 1 on GABA A R needs further investigation. Several techniques may help to illustrate the potential mechanism, such as molecular dynamics simulation and protein crystal structure analysis, and may provide the detailed protein structure of the GlyR␣ 1 -GABA A R complex and the potential binding sites of DH-CBD on the GlyR␣ 1 -GABA A R protein complex.

Cannabinoids restore GABA A R function in hyperekplexia
This study showed that DH-CBD had no direct effects on GABA A R in HEK293 cells. However, a previous report revealed that CBD and 2-arachidonoyl glycerol, an endogenous cannabinoid, can enhance the function of GABA A R in Xenopus oocytes (56). There may be two possible reasons for this contradiction. One may be the use of two distinct cell models in the two studies, amphibian oocytes and mammalian cell lines, which may carry quite different genomes, proteomes, and plasma membrane components (57,58). Another possible reason is that, although CBD and DH-CBD are very similar in structure, the subtle structural differences between them still lead to differences in their direct action on GABA A R.
To illustrate the therapeutic effects of DH-CBD on hyperekplexia disease, the R271Q site mutation was selected as for this study because, of all reported GlyR␣ 1 gene mutations, R271Q is the most common mutation causing hyperekplexia disease (21)(22)(23). In addition to R271Q mutant GlyR␣ 1 , GlyR␣ 1 carrying many other mutations, such as R218Q, P250T, V260M, S270T, and K276E, is also responsive to DH-CBD (25). These mutations, especially R271Q, have a high prevalence among all hyperekplexic patients. For example, in a clinical study, it was found that 10 of 17 hyperekplexic patients carried the GlyR␣ 1 R271Q mutation (23). However, several site mutations in the GlyR ␤ subunit and GlyT2 can also cause hyperekplexia (2-7).

Cannabinoids restore GABA A R function in hyperekplexia
Whether these hyperekplexia-causing mutations can affect the function of GABA A R remains unknown and needs to be investigated further. In addition, considering the fact revealed by this study that DH-CBD-induced functional restoration of GABA A R is based on GlyR␣ 1 , DH-CBD may not influence GABA A R under conditions involving mutations of the GlyR ␤ subunit and GlyT2 mutations. Thus, the therapeutic effect of DH-CBD may be based on gene sequencing results of individual hyperekplexic patients. For instance, DH-CBD or other cannabinoids may be an effective and precise medical treatment option for patients carrying GlyR␣ 1 mutations, especially Arg-271 mutations. However, for patients carrying GlyR ␤ subunitor GlyT2-related mutations, benzodiazepines may still be a preferential choice.

Ethics approval
All procedures were approved by the Institutional Animal Use and Care Committee of the School of Life Sciences, University of Science and Technology of China.

Site-directed mutagenesis
The R271Q, S296A, and K385A site mutations of ␣ 1 GlyR were introduced using the QuikChange Site-Directed Mutagenesis Kit (Takara, Inc.). The complementary DNA sequences were determined through dsDNA sequencing with a genetic analysis system (Sangon, Inc.).

Figure 4. Effects of GlyR␣ 1 S296A site mutation on DH-CBD-induced restoration on the interaction between GlyR␣ 1 and GABA A R.
A, representative trace records and average values of I Gly activated by 1 mM glycine in HEK293 cells coexpressing GABA A R (␣ 1 ␤ 2 ␥ 2 ) and GlyR␣ 1 carrying both R271Q and S296A mutation with or without 10 M DH-CBD preincubation (n ϭ 10). B, representative trace records and average values of I GABA activated by 1 mM GABA in HEK293 cells coexpressing GABA A R (␣ 1 ␤ 2 ␥ 2 ) and GlyR␣ 1 carrying the R271Q and S296A mutations with or without 10 M DH-CBD preincubation (n ϭ 13, 15, 9, 14, and 11). C, GlyR␣ 1 protein was purified using GABA A R ␣ 1 antibodies in HEK293 cells coexpressing GABA A R (␣ 1 ␤ 2 ␥ 2 ) and GlyR␣ 1 carrying the R271Q and S296A mutations with or without 10 M DH-CBD preincubation. Input represents the same protein immunoblots (IB) extracted from cell lysates prior to co-IP. Shown is quantification of WT and S296A mutant GlyR␣ 1 R271Q subunits binding to GABA A R ␣ 1 subunits with or without 10 M DH-CBD preincubation (n ϭ 4). The data were normalized to the WT group without DH-CBD preincubation. D, GlyR␣ 1 protein was purified using GABA A R ␣ 1 antibodies in HEK293 cells coexpressing GABA A R (␣ 1 ␤ 2 ␥ 2 ) and GlyR␣ 1 carrying the R271Q and S296A mutations with or without 10 M diazepam preincubation. Input represents the same protein immunoblots extracted from cell lysates prior to co-IP. Shown is quantification of WT and S296A mutant GlyR␣ 1 R271Q subunits binding to GABA A R ␣ 1 subunits with or without 10 M diazepam preincubation (n ϭ 3). The data were normalized to the WT group without diazepam preincubation. Data are represented as mean Ϯ S.D. *, p Ͻ 0.05; ***, p Ͻ 0.001; based on unpaired t tests; ns, not significant (p Ͼ 0.05).

Cannabinoids restore GABA A R function in hyperekplexia Coimmunoprecipitation
Plasmids coding GlyR␣ 1 and GABA A R were cotransfected into HEK293 cells using Lipofectamine 2000 reagent (Invitrogen). DH-CBD and equivalent vehicle were added separately to the medium 30 min before the cells were collected. 48 h after transfection, the cells were collected and treated with buffer containing 1 M Tris-HCl (pH 7.5), 1% protease inhibitor mixture (Roche), 1 M NaCl, and 5% sodium deoxycholate. To confirm the protein expression level, 60 l of whole-cell lysate was collected as the input before immunoprecipitation. The input has always been regarded as a standard in coimmunoprecipitation experiments. Then the primary antibody against GABA A R ␣ 1 protein was incubated with IgG-agarose beads overnight at 4°C. The mixtures were collected and incubated with the remaining cell lysates overnight. After five washes with cell lysis buffer, 100 l of loading buffer was added, followed by 5 min of boiling. The samples were then used for Western blotting. Samples were loaded on SDS-PAGE gels (12%) and then transferred to a PVDF membrane (NEN, Boston, MA) for 90 min. Then the membrane was blocked with Tris-buffered saline containing 0.1% Tween 20 and 5% (w/v) nonfat milk for 1 h. Primary antibodies against GABA A R ␣ 1 (1:100, 06-868, Merck), GlyR ␣ 1 (1:500, NB300-113, Novus), and GAPDH (1:5000, 60004-1-AP, Proteintech) were used for overnight incubation. After three 5-min washes, the membrane was incubated with secondary antibodies against rabbit (1:5000, ab6721, Abcam) for 2 h at room temperature. The membrane was washed three times, and the protein bands were imaged using ECL reagent (Thermo Fisher Scientific). The gray values were analyzed using ImageJ software (National Institutes of Health).

R271Q
, and YFP-GABA A R␣ 1 plasmids were created by inserting complementary DNA sequences coding CFP and YFP at the N terminus of GlyR␣ 1 and GABA A R␣ 1 in vector pcDNA3.1ϩ. For live imaging, cells were plated on 18 ϫ 18 mm glass coverslips (CITOGLAS) coated with poly-D-lysine (Sigma-Aldrich). The coverslips were mounted in custom-designed chambers using L-15 medium without phenol red (Thermo Fisher). The temperature was maintained at ϳ37°C using an air stream incubator. The CFP-GlyR␣ 1 WT , CFP-GlyR␣ 1

R271Q
, and YFP-GABA A R␣ 1 plasmids were then transfected equivalently using Lipofectamine 2000 according to the manufacturer's instructions and used for analysis 48 h later. The equivalent vehicle and DH-CBD were added separately to the transfected cells 2 h before live imaging. DH-CBD at 10 M was used in this experiment. The concentration of vehicle (ethanol) was less than 0.1%. The cells were observed using a Nikon Ti-Eclipse inverted microscope equipped with a charge-coupled device camera (Andor), a spinning disk confocal microscope (Yokogawa), and a laser merge module equipped with 445-, 488-, and 594-nm lasers (ILE, Andor). Fluorescence images were collected using iQ3 software (Andor). CFP and YFP were excited at 445 and 488 nm, respectively. CFP and YFP emissions were acquired simultaneously with a beam splitter (OPTOSOLIT II). The YFP/ CFP emission ratio in each image was calculated after background subtraction by MATLAB and averaged over multiple cells. Exper-iments were repeated multiple times with similar results. The fluorescence intensity was analyzed using ImageJ software.

Drugs
All chemicals, including diazepam, glycine, and GABA, were from Sigma-Aldrich. The external solution was prepared the day before the experiment. Before electrophysiological recordings, the agonists, modulators, and antagonists were diluted with external solution. DH-CBD and CBD were synthesized according to a procedure described previously (50). DH-CBD was dissolved with ethanol and diluted by external solution before recording.

Statistical analysis
Transfected HEK293 cells were picked randomly for electrophysiological experiments. Concentration-response data analysis was performed using the nonlinear curve fitting program. Data were fit using the following Hill equation: I/I max ϭ bottom ϩ (top Ϫ bottom)/(1 ϩ 10 (logEC 50 Ϫ log[agonist]) ϫ Hill slope). I max is the maximum current. Data were statistically analyzed by unpaired t tests using GraphPad Prism 6.0 (GraphPad Software). Data are presented as mean Ϯ S.D. p Ͻ 0.05 was considered significant.