2.1 Subjects and tissue processing

Brain tissue samples from nine patients who had undergone neurosurgery for refractory epilepsy were obtained from Hua Shan Hospital (Fudan University in Shanghai, China). All procedures and experiments in this study were approved by the Ethics Committee of Fudan University (Ethics No. KY2015-159). Informed consent was obtained from each patient for the use of brain tissue and access to medical records for research purposes. We collected core lesions and perilesional tissues from these  patients simultaneously for comparison using a standard procedure. Briefly, the origin of seizure and core lesion was determined by preoperative seizure monitoring, imaging, and electroencephalographic assessment. Patients were asked to discontinue the administration of anti-seizure medicines for at least three seizures to monitor the origin of seizures through video electroencephalography or stereoelectroencephalography. Simultaneously, PET/MR scans were conducted on the patients and a comprehensive determination of the core area was thus made in combination with the results from EEG monitoring. For patients with a clearly identified origin who were suitable for epilepsy surgery, three-dimensional reconstruction was conducted to plan the surgical scope of the lesion. During the surgery, navigation systems were utilized to achieve the resection of the lesion area. The scope of surgical resection will extend 1–2 cm beyond the nearest electrodes where no discharges were detected. The tissue located beyond the electrodes where no discharges were detected was defined as the perifocal tissue. Specimens were immediately stored in liquid nitrogen and maintained at −80°C (for western blotting (WB) and RT-qPCR) or fixed in 4% Paraformaldehyde fixative (for immunohistochemistry). Patient information is listed in Table S1.

2.2 Reagents and plasmid

Reagents of PTZ, CTZ, and SHP-1 inhibitor SSG were purchased from Sigma, USA. SHR-1210, the blocker of PD-1was a kind present sent by Shanghai Hengrui Pharmaceutical CO., LTD. Nivolumab was purchased from BMS, USA. The approximately 60 bp of the pre-miRNA targeting PD-1 mRNA sequence and non-effective scrambled pre-miRNA construct in a phSyn-EGFP-Mir155-SV40-puromycin vector were used to acutely silence PD-1 expression and synthesized by Genechem Co., Ltd. (Shanghai, China).

PD-1 pre-miRNA (5′-3′): TGAAGGTGGCATTTGCTCCCTGTTTTGGCCACTGACTGACAGGGAGCATGCCACCTTCACAGG.

Scrambled pre-miRNA (5′-3′): ACGTGACACGTTCGGAGAAGTTTTGGCCACTGACTGACTTCTCCGAGTGTCACGTCAGG.

2.3 Animals

Behavioral studies and electroencephalographic (EEG) recording were performed on male C57BL/6 mice weighing 22–26 g and male SD rats weighing 180–220 g. These animals were obtained from the Shanghai Experimental Animal Center of the Chinese Academy of Sciences. All the animals were maintained in an air-conditioned room with a suitable temperature at 22 ± 1°C under a 12 h light/dark cycle (7:00 a.m. to 7:00 p.m., lights on), and food and water were available ad libitum. All the experimental procedures were approved by the Committee of Animal Use for Research and Education of Fudan University (Ethics No. Y2019-053) and carried out in accordance with the Chinese National Science Foundation Animal Research Regulations. In the epileptic behavior experiments, we made all our efforts to minimize the number of animals used and their suffering, in accordance with the ethical guidelines for animal research.

2.4 Animal surgery procedure

Lateral ventricle cannula implantation: The animals were anesthetized with isoflurane and mounted in a stereotaxic apparatus with body temperature maintained at 37°C. The scalp was cut, and the skull was exposed. A little bur hole was made, with the tip of the guide cannula (26-gauge, 2.0 mm) placed above the left lateral ventricle in mice (AP −0.5 mm, ML −1.0 mm, DP −2.0 mm). Similarly, the guide cannula (22-gauge, 4 mm) for rat was placed above the left lateral ventricle in rat (AP −0.3 mm, ML −1.03 mm, DP −4.0 mm). The guide cannula was affixed to the skull with dental acrylic. The injection cannula was 30 gauge, 2.2 mm for mice and 27 gauge, 5.0 mm for rat.

Electrode implantation surgery: Two bur holes were drilled with two little stainless-steel screws (2-mm diameter and 2-mm length) embedded, one serving as a recording electrode in the right skull above the hippocampus of mice (AP −1.8 mm and ML 1.4 mm) or rats (AP −3.8 mm and ML 2.0 mm) and the other as reference electrode above the forehead of mice (AP 1.5 mm and ML −1.5 mm) or rats (AP 2.5 mm and ML −2.5 mm). Two screws were then connected to a connector plug with wires for later connecting to recording leads. Then, screws and connector plug were affixed to the skull with dental acrylic. After surgery, animals were kept for recovery for at least 5 days before the experiments.

2.5 Immunofluorescence

Fixed brain tissues were gradiently dehydrated using 15% and 30% sucrose in PBS, embedded in OCT, and frozen at −80°C before sectioning. A total of 40-μm thick slides were sectioned using a freezing microtome and then blocked in a blocking buffer consisting of 4% donkey serum and 0.25% Triton X-100 for 2 h at room temperature. Floating slides were incubated overnight at 4°C in primary antibodies Guinea Pig anti-NeuN (Millipore, ABN90P, 1:800), Mouse anti-PD-1 (Proteintech, 66220-1-Ig, 1:500), and Rabbit anti-Nav1.6 (Alomone Labs, ASC-009, 1:200) diluted in blocking buffer. Following 3 washes in PBS, sections were then incubated with secondary antibodies Alexa Fluor® 647 anti-Guinea IgG (Invitrogen, A-21450, 1:800), Alexa Fluor® 594 anti-Mouse IgG (Invitrogen, A-11005, 1:800), and Alexa Fluor® 488 anti-Rabbit IgG (Invitrogen, A-11034, 1:800) for 2 h at room temperature. Then, after washing in PBS for 3 times, sections were incubated with DAPI (Sigma, D9542, 1:1000) diluted in PBS for 15 min before being sealed with ProLong@Diamond antifade reagent (Invitrogen, P36965) and glass covers. Images were obtained using a fluorescence microscope (Nikon AIR).

2.6 Western blotting

Western blot analysis was performed to compare the PD-1 level in the core lesions and the perilesional tissues. Eight pairs of tissue samples were cut into small pieces, homogenized in lysis buffer (BioVision, CA95035) including protease inhibitors (Roche, 04906837001, 05892970001), and centrifuged at 10,000 × rpm at 4°C for 15 min. The protein concentration of the lysates was determined using a BCA protein assay kit (Thermo Fisher, 23227). The extracts (10 μg) were resolved by 10% SDS-polyacrylamide gel electrophoresis and electro-transferred to a polyvinylidene difluoride (PVDF) membrane, which was blocked with 5% skimmed milk in 0.5‰ TBST then incubated for 2 h at room temperature. Then, the PVDF membranes were incubated with primary antibodies rabbit anti-PD-1 (Proteintech, 18,106-1-AP, 1:500), mouse anti-GAPDH (Proteintech, 60004-1-Ig, 1:3000) in 5% skimmed milk overnight at 4°C, followed by incubation with a horseradish peroxidase-conjugate second antibodies for 2 h at 37°C. Chemiluminescent signals were generated by using SuperSignal West Femto maximum sensitivity substrate kit (Thermo Fisher, 34095), and the immunoreactive bands were visualized by enhanced chemiluminescence (ProteinSimple).

2.7 Quantitative real-time PCR

Total RNA was extracted from the seven pairs of specimens by using MiniBEST Universal RNA Extraction Kit (Takara, 9767). 1 μg of RNA was reverse transcribed to cDNA using oligo (dT) primer (Takara, RR600B). PCR primers were designed based on NCBI and synthesized by Sangon Biotech (Shanghai, China). The following primers were used: human-GAPDH, rat-GAPDH, human-PDCD1, and rat-PDCD1. The cycling conditions were as follows: Initial denaturation step was performed at 94°C for 1 min, followed by 40 cycles of denaturation step at 94°C for 30 s, annealing step at 55°C for 30 s, extension step at 72°C for 1 min, and finally the final extension at 72°C for 5 min. Relative quantification was calculated using the 2^−ΔΔCt method.

Human-PDCD1 (forward: 5′-CTTCACCTGCAGCTTCTCCA-3′; reverse: 5′-CCGCAGGCTCTCTTTGATCT-3′), rat-PDCD1 (forward: 5′-GATATCCCAGACCCTCACCCA-3′; reverse: 5′-CAGCTTCTCTGGCCTCTGACAT-3′).

2.8 Behavioral analysis and video EEG recording in animals

The male C57BL/6 mice and SD rats were randomly divided into saline control group and SHR-1210 (10 μg) pretreatment group (10 mg SHR-1210 dissolved in 1000 μL saline). Before intracerebroventricular injection with either saline or SHR-120, the animals were placed in the plastic cages and acclimatized for at least 30 min. Thirty minutes after intracerebroventricular injection, PTZ (50 mg/kg) was injected intraperitoneally to induce seizures. Epileptic behavior and video EEG were simultaneously recorded for 30–60 min after PTZ administration. For EEG recordings, electrophysiological signals were amplified (×1000) and sampled at 1 kHz (low-pass filtered at 100 Hz) using NeuroLog System (Digitimer Ltd, Heartford Shire, UK), digitized with CED Micro 1401 (Cambridge Electronic Design, Cambridge, UK), and recorded in a computer using Spike software (version 6.0, Cambridge Electronic Design, Cambridge, UK).

Epileptic behaviors in rodents were scored using the improved five-graded Racine Score system. Briefly, Racine score 1: facial clonus, head nodding; Racine score 2: tail erection, unilateral hindlimb clonus; Racine score 3: unilateral forelimb clonus, back clonus; Racine score 4: rearing with bilateral forelimb clonus; and Racine score 5: rearing and falling, loss of postural control. The assessment of the seizure severity in an animal was based on the maximal grade observed and the percentage of mice that developed generalized tonic–clonic seizures (GTCs).

2.9 Culture of the primary hippocampal cell and transfection

Primary hippocampal neurons were prepared from embryonic day 17–18 Sprague–Dawley rats as we previously reported.²⁸ Briefly, the pregnant rat was anesthetized with Urethane (25%, 0.4 mL/kg, i.p.) and the pups were dissected out for tissue preparation. After the dissection of the hippocampus, the tissue was rinsed in cold HBSS and then digested with 0.05% trypsin–ethylenediaminetetraacetic acid for about 20 min at 37°C. Then, DMEM with 10% FBS was used to end the digesting process. Single neurons were collected in the plating medium DF12 (DMEM, 10% FBS, 10% F12) and then planted on the coverslips, which were pre-coated by poly-D-lysine (0.1 mg/mL). After culturing for 1 day, half of the media were changed into neuronal culture media (neurobasal media containing 2 mM GlutaMAX Supplement, 2% B27, and 25 μg/mL penicillin/streptomycin). Ara-C (2 μM) was added 6–8 days after plating, and cells were fed twice weekly thereafter. All cells were grown at 37°C and in 5% CO₂. All culture reagents were ordered from Invitrogen. Hippocampal neuronal cultures were transfected with scram-EGFP or miPDCD1-EGFP plasmids using the calcium phosphate cell transfection protocol described previously.²⁸

2.10 Acute slice preparation

Male C57BL/6 mice, aged P28–35, were anesthetized with isoflurane. After decapitation, brains were dissected quickly and placed in ice-cold NMDG cutting solution containing (in mM) 92 NMDG, 82 NaCl, 2.5 KCl, 1.2 NaH₂PO₄, 20 HEPES, 30 NaHCO₃, 25 Glucose, 5 Na-ascorbate, 3 Na-pyruvate, 2 Thicurea, 10 MgSO₄, 0.5 CaCl₂, and pH 7.3–7.4 (300–310 mOsM). Slices (250 μm) were collected with a vibratome (Leica Instruments) and maintained in continuously oxygenated artificial cerebrospinal fluid (ACSF) for at least 30 min at 34°C and then at room temperature for 30 min before recording. Individual slices were transferred to a submerged recording chamber visualized with infrared optics microscope (Nikon), where they were perfused continuously with carbogen-buffered ACSF.

2.11 Whole-cell electrophysiology

In the experiment, standard whole-cell patch-clamp recordings were performed using a Multiclamp 700B amplifier (Axon), a Digidata 1440A digitizer (Axon), and pCLAMP 10.3 software (Axon). To record epileptiform burst in primary hippocampal neurons or acquire resting membrane potential (RMP) and input–output (I-O) curves of hippocampal neurons in CA1, the recording pipettes were pulled from borosilicate glass by pipette puller P-97 (Sutter Instrument), with the resistance of 3–5 MΩ when filled with intracellular recording solution containing (in mM): 125 K-gluconate, 10 KCl, 10 Hepes, 10 Tris-phosphocreatine, 4 Mg-ATP, 0.5 Na-GTP. pH was adjusted to 7.2 with 0.5 M KOH and the osmolarity was adjusted at ∼305 mOsm. The bath solution for cultured neurons contained (in mM): 128 NaCl, 30 D-glucose, 25 HEPES, 5 KCl, 2 CaCl₂, 1 MgCl₂. pH was adjusted to 7.3 with 0.5 M NaOH and the osmolarity was adjusted at ∼310 mOsm. The definition of epileptiform bursting was at least five consecutive action potentials overlaying a large depolarization shift, and we considered neurons showing at least two epileptiform bursts within 10 min to be epileptiform neurons. RMP was recorded within the first 10 s after achieving whole-cell configuration in current-clamp mode, while the cell is at rest without any holding current. The action potential input–output curve was generated by injecting positive current in current-clamp mode from −40 to 250 pA at 10-pA increments for 1000 ms.

To record sodium currents, recording electrodes were filled with an intracellular solution containing (in mM) 110 CsCl, 5 NaCl, 3 MgCl₂, 1 CaCl₂, 3 EGTA and 40 HEPES at pH 7.4. The bath solution was consisted of (in mM) 100 NaCl, 40 TEA-Cl, 3 KCl, 1 MgCl₂, 1 CaCl₂, 10 D-glucose, 10 HEPES, 1 BaCl₂, 1 CsCl, 2 4-AP and 0.1 CdCl₂ bubbled with 95% O₂–5% CO₂ at pH 7.2. For steady-state activation, the peak currents were determined using 50-ms pulses from −80 mV to 30 mV in 10-mV steps from a holding potential of −70 mV at 5-s intervals. Conductance as a function of voltage was obtained from the current–voltage relationship and fitted by the Boltzmann function: G/G_max = 1/(1 + exp [V − V1/2)/k] to determine the voltage midpoint (V1/2). For steady-state inactivation, cells were held at −140 mV, and the test potential was from −140 to 20 mV for 500 ms at 10-mV increments. A second pulse to −10 mV for 50 ms was used to assess channel availability. The normalized current was plotted against voltage, and steady-state inactivation curves were also fitted with the Boltzmann equation I/I_max = 1/(1 + exp [V − V1/2)/k] to determine the voltage midpoint (V1/2).

2.12 Co-immunoprecipitation

Human brain tissues were lysed in the NP-40 lysis buffer containing PMSF (1 mM), inhibitor cocktail, and phosphatase inhibitor. Lysates were centrifuged at 15,000 g for 10 min, and the supernatant was transferred into new tubes. A total of 10 μg PD-1 or Nav1.6 antibodies were added to Protein G Dynabeads for cross-linking for 1 h, then mixed with the supernatant above, and incubated overnight at 4°C to form an antigen–antibody compound on a revolving rotor. After 3 washes using washing buffer, the compound was eluted and separated. The supernatant was mixed with 5× loading buffer and heated for 5 min at 95°C for PD-1 WB or 2× loading buffer without heating for Nav1.6 WB.

2.13 Statistical analyses

Most data in figures were presented as mean ± S.E.M., as indicated in the Figure legends. For the figures regarding the maximum Racine score in two groups, the median and interquartile range (IQR) was displayed. Statistical analysis of results was completed with Prism GraphPad 8.0. Biochemical, behavioral, and electrophysiology data were analyzed using paired Student's t-test, unpaired Student's t-test, one-way ANOVA (multiple groups), Mann–Whitney test (two groups of ordinal variables), or Fisher's exact test (two groups of two categorical variables). p < 0.05 was considered to indicate a statistically significant difference. Data are presented in the mean ± SEM. *p < 0.05, **p < 0.01, ***p < 0.001.

3.1 PD-1 expression is elevated in the lesion tissue of patients with epilepsy

To explore the association between PD-1 and epilepsy, we first confirmed that the PD-1 protein is expressed in human neurons by performing immunohistochemistry. We collected focal tissues responsible for seizure onset as foci (lesion cores) and surrounding surgically resected tissues (perifocal tissues) as controls from patients who underwent lesion resection due to drug-resistant epilepsy. Detailed information about these samples is listed in Table S1. Double staining results revealed that PD-1 was colocalized with a neuronal marker (Figure 1A) but was barely colocalized with astrocytic and microglial markers (Figure S1). Given the fact that there was neuronal PD-1 expression in human brain tissues, we then analyzed the protein expression and mRNA level of PD-1 in surgically resected specimens. Eight pairs of cortical tissues from patients with refractory epilepsy were used for WB, and seven pairs were used for RT–qPCR. As shown in Figure 1B, the protein expression of PD-1 was significantly increased (n = 8, p = 0.0288) in the lesion core compared with perifocal tissues. Consistent with the results obtained by WB, real-time qPCR showed a tendency for PD-1 mRNA levels to be increased in the lesion core compared with the perifocal control tissues (n = 7, p = 0.0802) (Figure 1C). The data presented here revealed upregulated expression of PD-1 in the pathological lesion core compared with perifocal tissues at both the mRNA and protein levels, suggesting that elevated expression of PD-1 might be implicated in the pathology of epilepsy.

3.2 Anti-PD-1 treatment has anti-epileptic effects both in vivo and in vitro

Based on the finding that neuronal PD-1 was abnormally upregulated in epileptic tissues, we wondered about the function of neuronal PD-1. To learn whether the function of PD-1 can affect seizure severity, we examined the effect of blocking PD-1 activity in vivo on electroencephalograms and the behavior of rodents with acute seizures induced by PTZ, a classical chemoconvulsant.^{29, 30} We quantified the percentage of animals that developed GTCs and the average highest seizure score to determine the severity of seizure. PD-1 was acutely blocked by intracerebroventricular injection of the PD-1 monoclonal antibody SHR-1210 (10 μg/1 μL). The raw EEG traces shown in Figure 2A,B and the results of power spectral analysis shown in Figure 2C are representative data from one pair of rats treated with or without the anti-PD-1 antibody. The histograms of the percentage of animals that reached the highest seizure stage and the average highest seizure score in either species, which are shown in Figure 2D–G indicate that anti-PD-1 treatment successfully prevented PTZ-induced seizures both in mice (vehicle: 4.6 ± 0.3, n = 10; SHR-1210: 2.2 ± 0.6, n = 14; p = 0.0014) and in rats (vehicle: 4.0 ± 0.4, n = 4; SHR-1210: 2.3 ± 0.3, n = 3; p = 0.0857).

Given the finding that PD-1 blockade could protect animals from seizures, combined with the fact that PD-1 can regulate the function of ion channels,^{14, 15, 17} we hypothesized that PD-1 signaling can alter the electrical activity of neurons, thus playing a potential role in the generation of seizures. Therefore, to assess whether the excitability of neurons is altered by PD-1 blockade, we performed patch-clamp recordings on cultured primary hippocampal neurons. The electrical activities of hippocampal neurons were assessed via whole-cell patch recording after coincubation with CTZ (5 μM, 48 h), a convulsant drug that induces robust epileptiform activity in neurons.³¹ This in vitro epileptic model was established and characterized in our previous studies.^31-33 The definition of an epileptiform burst was at least five consecutive action potentials overlaying a large (>10 mV) depolarization shift lasting at least 300 ms,³¹ and we considered neurons showing at least two of these epileptiform bursts within a 10 min recording period to be neurons showing epileptiform activity. Coincubation of cultured primary hippocampal neurons with CTZ led to the transformation of neurons to epileptiform bursting neurons with robust synchronized epileptiform burst discharges, consistent with as previous reports,^31-33 with an average frequency of 0.106 ± 0.022 Hz (n = 25, Figure 3A,B). To test whether anti-PD-1 treatment is capable of altering epileptiform activity in this in vitro model, we applied SHR-1210 (100 μg/mL), an anti-PD-1 antibody, by acute perfusion. After perfusion with ACSF containing SHR-1210, the frequency of neuronal epileptiform burst discharges from the same neuron was substantially reduced to 0.057 ± 0.013 Hz (n = 25, p = 0.0002) (Figure 3A,B). In addition, we compared the burst frequency before and after continuous perfusion of ACSF containing no drug and found that this treatment did not have the same rundown effect over the same period as drug application (Figure S2). Moreover, since the SHR-1210 used in the current study was a monoclonal antibody produced and gifted by Hengrui Pharmaceutical (China) for experimental use before it was marketed for clinical use, we also tested the effect of an FDA-approved human anti-PD-1 monoclonal antibody (nivolumab). As shown in Figure S3, the burst frequency after perfusion with nivolumab (100 μg/mL) was 0.013 ± 0.006 Hz, which represented a substantial decreased from that observed before nivolumab perfusion (0.071 ± 0.025 Hz) (n = 6, p = 0.0362).

To further demonstrate that PD-1 functional deficits can attenuate CTZ-induced neuronal bursting, we knocked down the expression of PD-1 mRNA in cultured primary hippocampal neurons by transfecting them with PDCD1-miRNA labeled with EGFP (miPDCD1-EGFP) or scram-EGFP, serving as a control group where scrambled microRNA (miRNA) tagged with an enhanced green fluorescent protein (EGFP) marker was used to eliminate interference from transfection and the operation. We first verified the efficiency of PD-1 knockdown by applying RT–qPCR and confirmed that the mRNA level was significantly reduced (scram: 100.0 ± 0.0%, n = 3 v.s. miPDCD1: 71.7 ± 6.2%, n = 3; p = 0.01) (Figure 3F). We found that after CTZ incubation, both the proportion of neurons showing epileptiform burst activity and the average epileptiform burst frequency were significantly decreased in neurons transfected with miPDCD1 compared with neurons transfected with the control scrambled-EGFP plasmid. Furthermore, 6/14 neurons in the miPDCD1 group and 12/15 in the scrambled group showed epileptiform burst activity (p = 0.03) (Figure 3D,E), and the burst frequency of all the neurons tested was 0.008 ± 0.003 Hz (n = 15) in the miPDCD1 group and 0.025 ± 0.005 Hz (n = 14) in the scram-EGFP group (p = 0.015) (Figure 3F). Considering these results combined with those reported above, we concluded that PD-1 in primary hippocampal neurons is functional and that both acute blockade of PD-1 by a monoclonal antibody and genetic knockdown of PD-1 can markedly alleviate neuronal epileptiform activity and decrease the burst frequency of action potentials directly.

3.3 PD-1 regulates sodium channel function in neurons showing epileptiform activity

Considering the protective effect of anti-PD-1 treatment on neuronal epileptiform bursts, we wondered what the underlying mechanism is. It is well accepted that sodium channels play an important role in initiating action potentials in neurons and their dysfunction can cause inherited epilepsy.³⁴ Additionally, it has been reported that PD-1 is a potential regulator of sodium currents.¹⁵ Therefore, to further investigate the mechanism by which SHR-1210 diminishes convulsant-induced overexcitation of hippocampal neurons, we examined the effects of PD-1 inhibition on the kinetics of voltage-activated sodium channels in hippocampal neurons. The steady-state activation kinetics of sodium channels were first tested under control conditions, upon treatment with CTZ and following SHR-1210 perfusion. As shown in Figure 4A, sodium currents were determined by applying the voltage-clamp technique; representative types of sodium currents observed during application of these three treatment protocols are illustrated. Analysis of current density versus voltage relationships from the averaged data showed that the peak current density in the CTZ coincubation group was significantly higher than that in the control group and that this change could be reversed by SHR-1210 perfusion (Figure 4B). Figure 4C presents plots of the relative conductance (G/G_max) versus membrane potential in these three groups. These data were fitted with the Boltzmann equation, and the results revealed that in neurons with epileptiform discharges, the membrane potential required to elicit opening of half of the sodium channels was significantly shifted from −29.7 ± 1.5 mV (n = 9) to −41.8 ± 2.4 mV (n = 11, p = 0.0047), suggesting that the CTZ-induced neurons are more easily excited than control neurons. The half-activation potential of CTZ + SHR-1210-treated neurons was −42.9 ± 2.8 mV (n = 11), which was not different than that of CTZ alone-treated neurons, indicating that treatment with the anti-PD-1 antibody SHR-1210 did not affect the opening status of overactivated sodium channels in neurons induced by CTZ (Figure 4D).

Thus, we further tested the steady-state inactivation kinetics of sodium channels; typical test-pulse currents for potentials between −140 and +20 mV are shown in Figure 4E. The data were fitted with the Boltzmann equation, and as illustrated in Figure 4F, we revealed a depolarizing shift of 15 mV for sodium channels in the CTZ-treated group compared with the control group; that is, the midpoint potential for the normalized current–voltage curve was −41.7 ± 1.9 mV in the CTZ group (n = 7) compared with −56.4 ± 1.2 mV (n = 7, p < 0.001) in the control group. It is quite plausible that the convulsant effect of CTZ on neurons might have partially been due to the loss of sodium channel blockage. However, the midpoint potential after the application of SHR-1210 was decreased to −48.1 ± 1.7 mV (n = 7, p = 0.0329), which significantly differed from the midpoint potential of the CTZ alone group, suggesting that blockade of PD-1 by SHR-1210 could reverse the epileptogenic effect of CTZ via sodium channel blockage (Figure 4G).

3.4 PD-1 and SHP-1 regulate neuronal excitability together

The ability of PD-1 to perform its functions is dependent on the recruitment of SHP-1.^{17, 35} Therefore, we hypothesized that PD-1 acts together with SHP-1 to regulate neuronal excitability. To address our hypothesis, we applied SSG, a validated inhibitor of SHP-1,^{15, 36, 37} to the intracellular solution when performing patch-clamp recordings on primary cultured hippocampal neurons. As expected, the CTZ-induced epileptiform burst frequency (0.151 ± 0.029 Hz, n = 8) was significantly decreased to 0.03 ± 0.007 Hz (n = 10; p < 0.001) by acute application of SSG (Figure 5A,B). We then examined the effects of SSG on the kinetics of voltage-activated sodium channels as described above. We determined the midpoint potential of steady-state inactivation and various test-pulse currents for potentials between −140 and +20 mV are shown in Figure 5C. After fitting the data with the Boltzmann equation (Figure 5D), it was clear that the midpoint potential of inactivation sodium currents in CTZ-induced neurons was significantly depolarized by approximately 10 mV (control: −56.4 ± 1.2 mV, n = 7 vs. CTZ: −47.1 ± 0.7 mV, n = 11; p < 0.001), and intracellular SSG application significantly reversed this depolarization (−52.3 ± 0.7 mV, n = 9; p < 0.001) (Figure 5E). In contrast, fitting the steady-state activation data with the Boltzmann equation revealed hyperpolarization of the half-activation potential by 10 mV in CTZ-induced neurons (40.8 ± 1.7 mV, n = 12) compared to the control group (−30.9 ± 1.8 mV, n = 9; p = 0.0061); however, similar to anti-PD-1 antibody treatment, intracellular SSG application did not alter this change in epilepsy-related sodium channel steady-state activation (41.7 ± 2.6 mV, n = 11) (Figure 5F).

Furthermore, we found that acute blockade of SHP-1 by intracerebroventricular injection of SSG successfully decreased the incidence of GTCs and the average highest seizure score following PTZ (50 mg/kg, i.p.) injection in mice (Figure 5G,H). The average highest seizure score evoked by PTZ was 4.6 ± 0.3 (n = 14) in the control group compared to 1.6 ± 0.6 (n = 8) in the SSG group (p < 0.001) (Figure 5H). The similar effects of pharmacological blockade of PD-1 and that of SHP-1 on seizure severity with suggested that PD-1 and SHP-1 likely act together to regulate seizure activity.

3.5 PD-1 is coexpressed with SHP-1 and Nav1.6

Given that blockade of either PD-1 or SHP-1 impacted the kinetics of sodium channels, it is conceivable that there is an association between the PD-1 complex and sodium channels. Great strides have been made in understanding the structure of PD-1, which is a type I transmembrane glycoprotein consisting of a single immunoglobulin variable region (IgV)-like domain in the N-terminus, a transmembrane domain, and a cytoplasmic tail containing tyrosine-based signaling motifs.³⁸ Due to its structure and its regulatory effect on sodium channel kinetics, PD-1 reminds us of VGSC β subunits. VGSC β subunits are also type I transmembrane glycoproteins featuring an extracellular N-terminal domain with homology to V-type immunoglobulin (Ig) loops, a transmembrane segment, and a C-terminal tail (Figure 6A).³⁹ All these domains of VGSC β subunits are capable of interacting with VGSC α subunits and modulating α subunit gating and cell surface expression.⁴⁰ Inspired by the idea that PD-1 might act as a potential auxiliary subunit of VGSCs, we next investigated whether there is a structural interaction between PD-1 and VGSC α subunits.

There are nine genes encoding α-subunits, among which Nav1.1, Nav1.2, and Na1.6 are prevalent in the CNS.⁴¹ Nav1.6 is the dominant sodium channel in the adult brain and is widely distributed on the axon initial segment (AIS) of principal excitatory neurons and inhibitory interneurons, which is responsible for the initiation and propagation of action potentials.⁴² In addition, Nav1.6 has been found to be strongly associated with the onset of seizure activities.^43-45 Therefore, we utilized Nav1.6 as our major target and examined the interaction between PD-1 and Nav1.6 by performing immunohistochemistry and co-IP. We coimmunostained for Nav1.6 and PD-1 in tissues from patients with refractory epilepsy. Immunofluorescence and confocal imaging revealed PD-1 puncta and a high degree of colocalization between Nav1.6 and PD-1 on human neurons (Figure 6B,C). The results of colocalization analysis of PD-1 and Nav1.6 by immunofluorescence are shown in the scatterplot in Figure 6D, and Pearson's correlation coefficient (PPC) was 0.5 ± 0.05 (n = 6), suggesting fairly strong colocalization of PD-1 with Nav1.6 (Figure 6E). Interestingly, colocalization of Nav1.6 with PD-1 in the AIS could be seen and is indicated by the white arrows in Figure 6B. Moreover, a co-IP experiment also demonstrated that PD-1 was coprecipitated by an anti-Nav1.6 antibody in human brain lysates (Figure 6F). Moreover, we performed co-IP to validate that there was an interaction between PD-1 and SHP-1 (Figure 5G). The results indicated an extremely high possibility that there is a close interaction between the PD-1 complex and Nav1.6 in human brain tissue from patients with epilepsy. Therefore, it is possible that PD-1 impacts the inactivation kinetics of sodium channels via its C-terminal tail, thus playing a vital role in regulating neuronal excitability.

FUNDING INFORMARION

This work was funded by NSFC grants (81971204, 32111530119, 31771188) and supported by Shanghai Municipal Science and Technology Major Project (No. 2018SHZDZX01), ZJ Lab, and Shanghai Center for Brain Science and Brain-Inspired Technology. It also partly supported by a grant from Shenzhen Science and Technology Innovation Commission Municipality (JCYJ20210324103409023) to LW and a grant from the Sanming Project of Medicine in Shenzhen (No. SZSM202111010) to YW.