Proline‐rich transmembrane protein 2 specifically binds to GluA1 but has no effect on AMPA receptor‐mediated synaptic transmission

Abstract Background Proline‐rich transmembrane protein 2 (PRRT2) is a neuron‐specific protein associated with seizures, dyskinesia, and intelligence deficit. Previous studies indicate that PRRT2 regulates neurotransmitter release from presynaptic membranes. However, PRRT2 can also bind AMPA‐type glutamate receptors (AMPARs), but its postsynaptic functions remain unclear. Methods and results Whole‐exome sequencing used to diagnose a patient with mental retardation identified a nonsense mutation in the PRRT2 gene (c.649C>T; p.R217X). To understand the pathology of the mutant, we cloned mouse Prrt2 cDNA and inserted a premature stop mutation at Arg223, the corresponding site of Arg217 in human PRRT2. In mouse hippocampal tissues, Prrt2 interacted with GluA1/A2 AMPAR heteromers but not GluA2/A3s, via binding to GluA1. Additionally, Prrt2 suppressed GluA1 expression and localization on cell membranes of HEK 293T cells. However, when Prrt2 was overexpressed in individual hippocampal neurons using in utero electroporation, AMPAR‐mediated synaptic transmission was unaffected. Deletion of Prrt2 with the CRIPR/Cas9 technique did not affect AMPAR‐mediated synaptic transmission. Furthermore, deletion or overexpression of Prrt2 did not affect GluA1 expression and distribution in primary neuronal culture. Conclusions The postsynaptic functions of Prrt2 demonstrate that Prrt2 specifically interacts with the AMPAR subunit GluA1 but does not regulate AMPAR‐mediated synaptic transmission. Therefore, our study experimentally excluded a postsynaptic regulatory mechanism of Prrt2. The pathology of PRRT2 variants in humans likely originates from defects in neurotransmitter release from the presynaptic membrane as suggested by recent studies.


| INTRODUC TI ON
Since its first pathogenic mutation was identified in humans in 2011, proline-rich transmembrane protein 2 (PRRT2) (chromosome 16p11.2) has been considered the causative gene for several neurologic diseases, such as paroxysmal kinesigenic choreoathetosis, benign familial infantile epilepsy, and familial infantile convulsions with paroxysmal choreoathetosis. 1-3 PRRT2, which is selectively expressed in neurons and located at synapses, plays a crucial role in neuronal migration, spinogenesis, and synapse formation and maintenance during development. Previously, researchers have revealed its interaction with SNAP25, a plasma membrane SNARE protein, and Syts1 and 2, Ca 2+ sensors that mediate neurotransmitter release, suggesting that PRRT2 comprises a substantial component of the neurotransmitter release machinery at presynaptic terminals. 4,5 Although most studies have focused on PRRT2 function at the presynaptic membrane, 6,7 scattered evidence suggests that PRRT2 may regulate glutamate receptor function at the postsynaptic membrane. Prrt2 has been detected in postsynaptic densities in rodents, although at lower levels than found in presynaptic densities. 4 In 2014, Schwenk et al. reported a list of dozens of proteins that bind native α-amino-3-hydroxy-5-methyl-4-isoxazole-propionicacid receptors (AMPARs), which included Prrt2. 8 Subsequently, interaction of Prrt2 with AMPARs was verified in vitro and in vivo. 9 Here, we report the case of a patient carrying a PRRT2 mutant (c.649C>T; p.R217X) who had clinical manifestations of mental retardation. Transfection of Prrt2_R223X, a mimicking mutant from mouse Prrt2, in HEK 293T cells showed that the mutation led to the loss of Prrt2 protein. We analyzed the effects of Prrt2 on AMPARs with in vivo and in vitro systems and found that Prrt2 specifically bound to GluA1 but not the GluA2 or GluA3 subunits. In HEK 293T cells, Prrt2 suppressed the total protein level and localization of cotransfected GluA1. However, in hippocampal CA1 neurons, neither overexpression nor deletion of Prrt2 affected GluA1 expression or synaptic AMPAR function. Thus, we conclude that Prrt2 does not regulate AMPAR function in vivo.

| Animals
All experiments were performed in accordance with established protocols (certificate number: AP#SY06) approved by the Institutional Animal Care and Use Committees of Nanjing University. Three litters of C57BL/6JGpt mice (age, 0-42 days) were used for age-dependent Prrt2 expression analysis. Six female ICR mice (age, 4-6 weeks; weight, 50-60g) were utilized for in utero electroporation (IUE).
Five Cas9-knock-in (B6/JGpt-Rosa26 tm1(CAG−Cas9−tdTomato) /Gpt) mice and 6 C57BL/6JGpt mice (age, 0 days) were used for primary neuron culture. All mice were purchased from the Model Animal Research Center (Nanjing University). Mice were housed in pathogen-free conditions at 22°C, 55% relative humidity, and under a 12-h light/ dark cycle, with provision of food and water ab libitum.

| Experimental constructs
The cDNAs of mouse GluA1, GluA2, GluA3, and Prrt2 were used in this study. The HA-tagged and FLAG-tagged recombinant proteins used for Western blotting were generated by overlapping PCR (Vazyme Biotech, P505) and subcloned into the pCAGGS vector.
An HA-tag was attached to the N-terminals of GluA1, GluA2, and GluA3 and was used for Western blot detection of these proteins.
To generate the Prrt2 c.667C>T (p.R223X) construct, we performed site-directed mutagenesis by PCR using the wild-type (WT) vector.
Mutant constructs were confirmed by sequencing over the entire length of the coding region.
Design and screening of single-guide (sg)RNAs for the clustered regularly interspaced short palindromic repeats (CRISPR) constructs were performed as previously described. 10 The Prrt2 sgRNA was designed to target part of the coding region of exon 2. The primers used were as follows: 5′-ACCGTTCAGCCGGGCCCAGGCATC-3′ (forward) and 5′-AAACGATGCCTGGGCCCGGCTGAA-3′ (reverse).
The sgRNA expression vector was constructed by inserting the in vitro synthesized PRRT2 sgRNA-targeted sequence into a vector that contained a tracrRNA sequence, and expression of the fused sgRNA was driven by the U6 promoter. After verifying the efficiency of the system, spCas9 was subcloned into the preceding vector.

| Co-immunoprecipitation
HEK 293T cells were co-transfected with the indicated expression plasmids in 10-cm dishes 48 h before use. Cells were washed three times with phosphate-buffered saline (PBS), harvested, and solubilized in co-immunoprecipitation assay lysis buffer (Bio TeKe Corporation, Shanghai, China) and 1 mM phenylmethylsulfonyl fluoride for 1 h at 4°C. After centrifugation at 13,800 × g for 20 min, the pellet was discarded. Lysates were then incubated with antibodies at 4°C overnight. Then, the lysates were incubated with Protein G beads (GE Healthcare, USA) for 2 h at 4°C on a rotating platform.

| In utero electroporation
Embryonic day 15 (E15) pregnant mice were anesthetized with 1% pentobarbital sodium (dissolved in normal saline) with 100 μl per 10 g mice dose by peritoneal injection before surgery. 11,12 To visualize the electroporating process, plasmids were mixed with 2 mg/ml Fast Green (Sigma-Aldrich), and pCAG-U6-sgRNA-hUbc-spCAC9-T2A-GFP was used at a final concentration of 2 μg/μl. During surgery, the uterine horns were exposed, and one lateral ventricle of each embryo was pressure injected with 1-2 μl of plasmid DNA.
Injections were performed by inserting a pulled glass microelectrode into the lateral ventricle through the uterine wall and embryonic membranes and injecting the content of the microelectrode by pressure. The embryos were then electroporated with five 50 ms, 40 V pulses delivered at 1 Hz using platinum Tweezertrodes with a square-wave pulse generator (BTX, Harvard Apparatus). Following electroporation, the embryos were placed back into the abdominal cavity, and the muscle and skin were sutured. The pregnant mice were then allowed to recover from surgery, and the pups were normally delivered. The full gestation period of each pregnant mouse is 19-21 days. All maternal mice that suffered with IUE were recovered from the surgery, and pups were delivered naturally. The maternal mice were humanely euthanized with CO 2 at the end of the nursing period. Death was confirmed by observing respiration and by using the corneal reflection method.

| Electrophysiology
Voltage-clamp recordings were performed on CA1 pyramidal neurons in acute hippocampal slices. The acute hippocampal slices were obtained from mice anesthetized (1% pentobarbital sodium) and decapitated at 21-28 days after IUE. To prepare acute slices, 300μm transverse slices were cut using a Leica vibratome (Leica Synaptic responses were evoked by stimulating the stratum radiatum of the CA1 region with a bipolar metal electrode. To ensure stable recording, the membrane holding current, input resistance, and pipette series resistance were monitored throughout the recording. Data were collected using a MultiClamp 700B amplifier (Axon Instruments, Molecular Devices), filtered at 2 kHz, and digitized at 10 kHz.

| Statistical analysis
Normalization was performed by dividing both the control and experimental conditions by the average value of the control. The paired whole-cell data were analyzed using the two-tailed Wilcoxon signed-rank test, and unpaired data using the Mann-Whitney U test. The one-way ANOVA test for multiple comparisons was used to analyze all the other experiments involving unpaired data. Data analysis was performed using Excel (Microsoft) and GraphPad Prism (GraphPad Software).

| Consent
Written informed consent to participate in this study was obtained from the patient and his parents.

| Case report of the PRRT2 c.649C>T mutant
The proband, a 14-year-old male patient from Jiangsu Province, was admitted to our center for genetic counseling and further evaluation in July 2017 with paroxysmal epilepsy and dysgnosia. The patient was delivered vaginally, and his first epilepsy event occurred at 5 months after birth, with normal brain computed tomography and electroencephalography results (Yancheng Hospital). At the age of  Figure 1C, D). The same mutation was also found in his mother.
Nevertheless, his mother, also a carrier of the heterozygous mutation, did not show similar symptoms.

| The mutation causes loss of protein expression
PRRT2 is highly conserved in mammals. To understand the pathology of the mutation found in our patient, we cloned mouse Prrt2 and introduced an R223X mutation to mimic that found in human PRRT2. A FLAG epitope was added to the C-termini of WT and mutant Prrt2 to facilitate protein detection. When transfected in HEK 293T cells, both the Prrt2 and FLAG signals were undetectable in the cells that expressed mutant Prrt2, in sharp contrast to cells that expressed WT Prrt2. These results suggest that the truncation mutation leads to loss of Prrt2 expression, consistent with a previous study reporting that truncated Prrt2 is unstable or not expressed 13 ( Figure 2A).

| Prrt2 specifically interacts with GluA1
To investigate the endogenous expression pattern of Prrt2 in various developmental stages, hippocampus tissues from mice on P0 to P42 were homogenized and incubated with a Prrt2 antibody. β-Tubulin, a housekeeping gene, was used as an internal control. Western blotting analysis indicated that Prrt2 expression gradually increased from a low level at birth and reached a plateau at P14 (Figure S1A,   B). Next, we examined the interaction between Prrt2 and AMPARs.
Co-immunoprecipitation experiments were performed with homogenates from the adult mouse hippocampus. We examined the ability of Prrt2 to bind to GluA1, GluA2, and GluA3 because they comprise the majority of AMPAR subunits in the hippocampus. 14,15 We found that both the GluA1 antibody and GluA2 antibody pulled down Prrt2 ( Figure 2B, C). Conversely, GluA3 did not interact with Prrt2 ( Figure 2D). Because AMPARs in the hippocampus mainly comprise heteromeric tetramers of GluA1/A2 or GluA2/A3, 15 these data indicated that GluA1/A2, but not GluA2/A3, interacts with Prrt2. If this prediction is correct, then pull-down of Prrt2 by GluA2 could occur via mediation of GluA1.
We then studied the interaction of Prrt2 with AMPAR subunits in HEK 293T cells. FLAG-tagged Prrt2 was co-expressed with GluA1, GluA2, and GluA3 tagged with an HA epitope at the N-terminus following the signal peptides. As predicted, the co-immunoprecipitation results showed that GluA1 interacted with Prrt2, while GluA2 or GluA3 did not interact ( Figure 2E). These results verified our prediction that, in hippocampal tissue, GluA2 would indirectly pull down Prrt2 via GluA1.

| Prrt2 suppresses GluA1 protein expression levels in vitro
We next examined the effects of Prrt2 on AMPAR expression. HAtagged GluA1 was co-transfected with WT and mutant Prrt2 into HEK 293T cells. After 3 days of expression, biotin was used to label surface proteins. HA signals from whole-cell homogenates and biotin-labeled membrane proteins were analyzed to determine the total and surface GluA1 content, respectively. Compared with the control group expressing HA-tagged GluA1 alone, Prrt2 suppressed total protein levels of GluA1, while co-transfection of Prrt2_R223X did not suppress GluA1 ( Figure 2F, G). Meanwhile, the surface expression level of GluA1 was also decreased after co-transfection with Prrt2, consistent with a previous report 9 ( Figure 2F, H). In contrast, co-expression of Prrt2 had no effect on the total and surface expression levels of GluA2 ( Figure S1C-E), consistent with the observation that Prrt2 specifically interacts with GluA1. These results demonstrated that Prrt2 suppresses GluA1 expression in vitro.

| Overexpression of Prrt2 does not affect synaptic AMPAR function
After characterization of the interaction of Prrt2 with AMPARs in vitro, we then studied its effects on synaptic AMPAR function. We

| Deletion of Prrt2 does not affect synaptic AMPAR function in neurons
There are two possible explanations for the lack of changes in synaptic function of AMPARs. One is that Prrt2 does not regulate AMPAR expression in neurons. Alternatively, the endogenous Prrt2 in we knocked out endogenous Prrt2 in hippocampal CA1 neurons using the CRISPR/Cas9 technique, which has been shown to efficiently delete targeting molecules in neurons. 16 We developed a Prrt2-knockout construct, CRISPR_Prrt2, containing both a Prrt2-targeting sgRNA and Cas9. The Cas9 cDNA was fused with GFP by T2A sequence so that GFP signal represents Cas9 expression ( Figure 4A). 17  in the test group was reduced by 70%, and the negative sgRNA exhibited no effect ( Figure S2D, E). In cultured neurons isolated from hippocampi of the Cas9-knock-in mice, lentivirus-mediated expression of Prrt2 sgRNA nearly completely depleted Prrt2 ( Figure S2F, G).
These results demonstrated that the sgRNA was highly effective in eliminating Prrt2. Then, the constructed CRISPR_Prrt2 plasmid was injected into ventricles of E15 mice, and hippocampal CA1 pyramidal neurons were transfected by IUE, as previously described ( Figure 3A).
Simultaneous dual whole-cell recordings from a transfected GFPpositive cell and a neighboring control neuron showed that deletion of Prrt2 had no obvious effect on the amplitude of AMPAR-EPSCs ( Figure 4B, C). Furthermore, the decay kinetics of AMPAR-EPSCs was also unaltered by Prrt2 deletion, indicating that the composition of synaptic AMPARs is not changed ( Figure 4D). Meanwhile, neither NMDAR-EPSCs nor the PPRs were altered by deletion of Prrt2 ( Figure 4E-G). These results demonstrate that Prrt2 deletion in mouse CA1 neurons does not affect synaptic trafficking of AMPARs.

| Overexpression or deletion of Prrt2 does not affect the surface/intracellular ratio of GluA1
Our electrophysiological analysis indicated that synaptic AMPARs are not altered by overexpression or deletion of Prrt2. Neurons  cells. 9 We found that antibodies against Prrt2 can pull down GluA1 and GluA2 but not GluA3 in mouse brain tissues. In HEK 293T cells, Therefore, we concluded that GluA2 pulled down Prrt2 indirectly through GluA1 in brain tissues. This finding is consistent with the notion that AMPARs in the cortex/hippocampus mainly exist in the GluA1/A2 and GluA2/A3 forms. 15 Recently, observation of native AMPARs using cryo-electron microscopy technology identified GluA1/A2/A3 type AMPARs in the brain. 20 However, antibodies against Prrt2 failed to pull down GluA3, indicating that either Prrt2 does not bind to GluA1/A2/A3 type AMPARs or that the amount of this type of AMPAR is minimal in the brain.
A previous study has verified that WT PRRT2, but not its truncated mutants, suppresses the surface distribution of GluA1 in vitro. 9 In the current study, we found that Prrt2 suppresses both total and surface GluA1 in HEK 293T cells, largely reconstituting previous observations. 9 However, overexpression or deletion of Prrt2 in hippocampal neuronal culture had no obvious effects on GluA1 expression or membrane distribution. Manipulation of Prrt2 in CA1 pyramidal neurons in vivo also did not affect AMPAR-EPSCs. AMPARs in hippocampal CA1 neurons are mostly GluA1/ A2 heteromers, which is a slow type of AMPARs. 15 If this GluA1/ A2 is replaced by faster AMPARs such as GluA2/A3 (another component of AMPARs in CA1 neurons), then the decay kinetics of AMPAR EPSC will be speeded. We thus calculated the decay kinetics of AMPAR-EPSCs and found no change ( Figure 3D, Figure 4D There is a possibility that even though Prrt2 does not regulate AMPAR function in rest condition, it may change activitydependent neuronal plasticity. We believe this is unlikely as any factor that has a role in neuronal plasticity, it generally affects basic transmission. 11,[21][22][23] There are several possible explanations as to why Prrt2 suppresses GluA1 in HEK 293T cells but not in AMPARs in neurons.
First, the expression of GluA1 in HEK 293T cells cannot fully mimic GluA1/A2 in neurons. Second, many other AMPAR binding proteins exist such as transmembrane AMPAR regulatory proteins and cornichons. 21,24 These factors may impose stronger regulation on AMPARs, which may overwhelm the effects of Prrt2. Third, neurons might have stronger regulatory capability than HEK 293T cells. For instance, it is assumed that Prrt2 facilitates GluA1 degradation in HEK 293T cells, while degradation of AMPARs is strongly controlled by factors other than Prrt2. Therefore, overexpression or deletion of Prrt2 produces little effect.
Overall, our study demonstrates that although it specifically binds to the GluA1 subunit, Prrt2 is not involved in regulating surface trafficking or basic transmission of AMPARs in hippocampal CA1 neurons. It would be of interest to learn whether Prrt2 exhibits postsynaptic regulation of AMPARs in interneurons or neurons in other brain regions. It should be noted that our conclusion was based on overexpression and deletion of mouse neurons. Whether this conclusion can be fully extend to human or the patient needs to be verified in the future.

CO N FLI C T O F I NTE R E S T
The authors declare no conflict of interest.