Kir4.1 channels in NG2-glia play a role in development, potassium signaling, and ischemia-related myelin loss

The contribution of the inwardly rectifying K+ channel subtype Kir4.1 has been focused mainly on astrocytes, where they play important roles in the maintenance of resting membrane potential, extracellular K+ uptake, and facilitation of glutamate uptake in the central nervous system. Here, we report the role of Kir4.1 channels in NG2-glia during brain development, potassium signaling, and in an ischemic stroke disease model. Kir4.1 channels are widely expressed in NG2-glia during brain development. In the adult mouse hippocampus, Kir4.1 channels in NG2-glia constitute more than 80% of K+ channels inward currents. This large portion of Kir4.1 channel currents exhibits a deficit in NG2-glia as an initial response in a transient ischemic mouse model. Further evidence indicates that Kir4.1 deficits in NG2-glia potentially cause axonal myelin loss in ischemia through the association with oligodendrocyte-specific protein (OSP/Claudin-11), which unravels a potential therapeutic target in the treatment of ischemic stroke.

T he inwardly rectifying K + channel subtype Kir4.1 has been well studied in astroglia within the central nervous system. Kir4.1 ion channels play prominent roles in the maintenance of resting membrane potential (RMP), extracellular K + uptake, cell volume regulation, and facilitation of glutamate uptake, as well as in neurodegenerative diseases [1][2][3][4][5][6] . Although a pioneer developmental study indicated that Kir4.1 could be immunoactivated in NG2+ glial cells in rat optic nerve 7 , NG2glia, which are known as oligodendrocyte precursor cells (OPCs), until recently were found to express high levels of Kcnj10 gene (which encodes Kir4.1) in juvenile mouse brain, as evidenced by RNA-Seq transcriptome analysis 8,9 . However, whether NG2-glia functionally express Kir4.1 channels in the adult brain, as well as their newly found physiologic and/or pathologic relevance are largely unexplored 10,11 . Different from astrocytes, NG2-glia demonstrate self-renewal functionality as multipotent stem cells by providing myelinating oligodendrocytes during early brain development and receive direct synaptic contacts from both glutamatergic and GABAergic neurons [12][13][14][15][16] , suggesting that NG2-glia have much closer interactions with local neurons and greater impact on neural networks.
Stroke is a neural disease clinically manifested by transient or permanent brain dysfunction symptoms. As one of the three most common diseases in the world, stroke has a high mortality and disability rate. Ischemic stroke is the most common form, accounting for 87% of strokes and mainly causes impairment of neural cells and ultimately the loss of brain function due to ischemia and hypoxia. To date, treatment options are still limited 17,18 . Although there have been reports that glial cells contribute to the stroke pathology, the causes, disease mechanisms, and potential impacts remain unclear 11,[19][20][21][22][23] . In the present study, we investigate the role of Kir4.1 channels in NG2-glia during brain development, potassium signaling, and in ischemiarelated myelin loss. We demonstrate Kir4.1 channels in NG2-glia are widely expressed during brain development and constitute the bulk of K + channel inward currents in adult hippocampus. Notably, in a transient middle cerebral artery occlusion (tMCAO) mouse model, Kir4.1 channel deficits in NG2-glia cause axonal myelin loss, thus unraveling a potential therapeutic target in the treatment of ischemic stroke.

Results
Expression of Kir4.1 in NG2-glia. To gain further insights into the role of Kir4.1 ion channels in NG2-glia, we undertook an investigation of these channels in a NG2DsRedBAC transgenic mouse strain. Fluorescent DsRed-labeled NG2-positive cells from postnatal 2-week-old mice were harvested and purified by fluorescence-activated cell sorting (FACS) (Fig. 1a, Supplementary Figure 1a). Both PCR and Western blot results illustrated that NG2-glia express Kir4.1 ion channel mRNA and protein (Fig. 1b), which is consistent with previous transcriptome and electrophysiological characterizations during early brain development 8,24 .
To investigate whether Kir4.1 channel expression persists during NG2-glia development, we further confirmed Kcnj10 gene expression in adult NG2-glia using combined RNA-seq transcriptome and single-cell reverse transcriptase polymerase chain reaction (RT-PCR) techniques in PDGFRα-creERT; ROSA26-mGFP transgenic mouse brain, as this mouse strain displays a very high efficiency in labeling NG2-glia during brain development (Fig. 1c, d, Supplementary Figure 1b) 25,26 . Whole-cell patch recordings from single GFP-positive NG2-glial cells in acute hippocampal slices from 2-to 3-month-old PDGFRαCre-mGFP transgenic mice clearly demonstrated macroscopic K + currents and Ba 2+ -sensitive currents 4 (Fig. 1e). To accurately define the percentage of Kir4.1 channel-dependent currents involved in total inward K + channel currents in adult NG2-glia, a tamoxifeninduced PDGFRαCre ER -mGFP; Kir4.1 f/f transgenic mouse was introduced to produce specific deletion of Kir4.1 in NG2-glia (Fig. 1f). Although no overt neurological deficits can be observed in Kir4.1 cKO mice at the examination time point P60, NG2-glia exhibited, on average, about 22.2 mV depolarization of the RMP and 7.5-fold increase of cell membrane resistance of GFP-labeled NG2-glia recorded in the hippocampal CA1 stratum radiatum (SR) compared with that in wild-type NG2-glia (Supplementary Figure 2). In addition, 82.2 ± 4.1% (n = 12) of Ba 2+ -sensitive currents in PDGFRα + NG2-glia at a holding voltage of −140 mV was eliminated in Kir4.1 cKO mice (Fig. 1g, h), which further confirmed that Kir4.1 channel current contributes to a large portion of inward K + channel currents in adult NG2-glia.

Discussion
In summary, our evidence strongly demonstrates that (1)  (2) Kir4.1 channel currents constitute a large portion of total inward K + channel currents in NG2-glia and an initial deficit of Kir4.1 channels in NG2-glia but not astrocytes occurred in a tMCAO mouse model; (3) deficits of Kir4.1 channels in NG2-glia contribute to the myelin loss of axons in tMCAO which is potentially through the association of Olig-specific protein (OSP/ Claudin-11). Our data systematically showed that NG2-glia functionally express Kir4.1 in adult brain and the impairment of myelin in axons was possibly caused by the deficits of Kir4.1 in NG2-glia after ischemia. NG2-glia, known as OPCs, sustain oligodendrocyte maturation, differentiation, and myelination during early brain development 2,13,14,16 . It has been reported that the K + channel family plays an important role in axon's myelination as well as in myelin-related brain damage [33][34][35][36] . Our data also support that K + channel subtype Kir4.1 which is expressed in NG2glia is crucial to the myelin formation during brain development and could contribute to the myelin loss after ischemia (Fig. 5). In normal brain, Kir4.1 upregulates in NG2-glia to sense local K + rises induced by neuronal activities 10 . However, in conditions of brain damage such as ischemia, both elevated [K + ] o and low pH exacerbate Kir4.1 channel openings in NG2-glia 30,31 , which in turn impairs OSP/Claudin-11 to cause a direct loss of myelin in axons, although an intrinsic mechanism of Kir4.1-OSP interaction remains unclear 35 . By providing a close spatial contact and synaptic interactions with surrounding neurons 10,12,15 , NG2-glia enable a fast response during normal healthy conditions and could exhibit higher vulnerability than other types of glial cells as well when triggered by hypoxia-induced ischemic signals. Therefore, NG2-glia not only play a role as cell reservoir for sustaining oligodendrocyte maturation and differentiation in the brain, they could also respond rapidly by directly interacting with neurons in myelination during the ischemic disease process. The observation that deficits of Kir4.1 channels in NG2-glia contribute to the loss of myelin in tMCAO highlights a new role of NG2-glia in physiological/pathological conditions in the brain and sheds light on a potential therapeutic target of NG2expressing Kir4.1 channels for the treatment of ischemic stroke. MCAO model in mice. The brain ischemic stroke mouse model was established as described before with slight modifications 38 . In brief, six-to eight-weekold male mice (~22 g) were anesthetized with an injection of 5% chloral hydrate (20 mL kg −1 ). Rectal and temporalis muscle temperature was maintained at 37 ± 0.5°C with a thermostatically controlled heating pad and lamp. Preparation of brain slices and electrophysiological recordings. For preparation of brain slices, mice were deeply anesthetized with 5% chloral hydrate and intracardially perfused with ice-cold carbogenated (95% O 2 , 5% CO 2 ) ACSF containing: 125 mM NaCl, 2.5 mM KCl, 1 mM MgCl 2 , 2 mM CaCl 2 , 1.25 mM NaH 2 PO 4 , 25 mM NaHCO 3 , and 12.5 mM D-glucose. Coronal sections of the brain were cut into 300 µm thickness (VT1200S; Leica Microsystems, Germany) and allowed to equilibrate for at least 1 h at 31˚C in ACSF which continuously bubbled with a mixture of 95% O 2 /5% CO 2 gas. Intracellular solutions for NG2-glial and astrocytic recordings consisted of (in mM) the following: 125 K-gluconate, 15 KCl, 8 NaCl, 10 HEPES, 0.2 EGTA, 3 Na 2 -ATP, and 0.3 Na-GTP and pH set to 7.3 (~305 mOsm). Slices were visualized with an upright epifluorescent microscope (BX51WI; Olympus, Tokyo, Japan) equipped with differential interference contrast optics and an infrared CCD camera (optiMOS, Q IMAGING; Olympus, Tokyo, Japan). Whole-cell recordings were made from NG2-glia and astrocytes in hippocampal CA1 stratum radiatum, with a MultiClamp 700B amplifier (Molecular Devices, Sunnyvale, CA, USA). Signals were low-pass filtered at 2 kHz and sampled at 20 kHz using Digidata 1550A (Molecular Devices) in all experiments.   d Bar graph shows 44%, 48%, and 58% decrease of total OSP protein levels in ipsilateral regions after tMCAO at different time checking points compared with the sham control. n represents the number of mice. Statistical significance was assessed as indicated using ANOVA followed by Dunnett Multiple Comparison tests with sham as control. e Representative images of CC1+ oligodendrocytes in both contralateral and ipsilateral cortex after tMCAO mice at postnatal 8 weeks. Scale bar: 20 µm. The bar graph on the right indicates that there was no difference of oligodendrocyte numbers between the infarction and contralateral region of tMCAO. The error bars represent s.e.m. n = 5 mice. P > 0.05 as indicated using two-tailed Mann-Whitney test. f The co-Immunoprecipitation results on the left panels show OSP reciprocally binding with Kir4.1 in WT mouse brain tissue. The panel on the right shows the reduction of Kir4.1 and OSP interactions in ipsilateral brain tissue of tMCAO mice at postnatal 8 weeks compared with its contralateral side when the same quantity of OSP protein is immunoprecipitated in these two lysates the Pierce Classic IP kit (Thermo Scientific Pierce, USA) according to the manufacturer's instructions. For co-immunoprecipitation, extracts from isolated brain tissues were precleared with Protein G-agarose at 4°C for 30 min. Then 4 µg of desired antibodies or control normal IgG was added to the lysates and incubated with protein G-agarose beads overnight at 4°C. The next day, the beads were washed with the IP lysis/Wash buffer three times. The immunoprecipitants were eluted using 2× Non-reducing Lane Marker Sample Buffer. All the immunoprecipitated samples were subjected to SDS Western blotting.

Methods
RNA and protein extraction from NG2-glia by FACS. Mice expressing DsRed under NG2-glia-specific CSPG4 promoter (NG2DsRedBAC) were used to purify NG2-glia by FACS. The brain tissues from NG2DsRed mice at P14 were dissociated following published guidelines 39 with slight modifications. Briefly, the brain tissues were dissected and digested for 90 min at 36˚C in 50 mL centrifuge tubes with 10 ml papain solution (1× EBSS, 0.46% glucose, 26 mM NaHCO 3 , 50 mM EDTA, 75 U mL −1 DNase I, 300 units of papain, 2 mM L-cysteine) bubbling with 5% CO 2 , 95% O 2 . After digestion, the tissue was washed four times with ovomucoid solution (1× EBSS, 0.46% D-glucose, 26 mM NaHCO 3 , 1 mg mL −1 ovomucoid, 1 mg mL −1 BSA, and 60 U mL −1 DNase I) and mechanically dissociated with two fire-polished borosilicate glass pipettes with different bore sizes. A bottom layer of concentrated ovomucoid solution (1× EBSS, 0.46% D-glucose, 26 mM NaHCO 3 , 1 mg mL −1 ovomucoid, 1 mg mL −1 BSA, and 60 U mL −1 DNase I) was added to the cell suspension. The tubes were centrifuged at room temperature at 300 g for 10 min, and the resultant pellet was re-suspended in D-PBS with 0.02 % BSA and 13 U mL −1 of DNase I, and filtered with a 40 µm mesh. FACS was performed in a BD FACSAria II Flow Cytomerter (BD Bioscience) with a 70 µm nozzle using standard methods at Shanghai Jiao Tong University, Core Facility of Basic Medical Sciences and analyzed with FlowJo software. For RNA extraction, sorted cells were collected in D-PBS with 0.1% BSA, and centrifuged for 10 min a`t 4°C and 2000 g. The RNA was extracted from the pelleted cells using Trizol reagent (Thermo Scientific, Pierce, USA). For protein extraction, cells were collected in D-PBS and, right after FACS, cells were incubated with lysis buffer (150 mM NaCl, 1% Triton X-100, 12 mM Na + -deoxycholate, 3.5 mM sodium dodecyl sulfate, 50 mM Tris pH 8, and Protease Inhibitor cocktail) at 4°C for 40 min.
Single-cell RT-PCR and RNA-sequencing. Single NG2-glia with GFP fluorescence labeling from PDGFRαCreER; mGFP mice at postnatal 7 weeks was selected and aspirated into a glass pipette from hippocampal acute slices following a method described previously with slight changes 40 . In brief, cells were picked promptly by micromanipulation and immediately placed in lysis buffer. To minimize the changes in gene expression and meet the quality requirement for cDNA used to construct sequencing libraries, all NG2-glial cells were collected within 3 h after slice preparation. The selected NG2-glia were processed for singlecell RNA extraction and reverse transcription within 1 h and were subjected to cDNA amplification and purification. Single-cell cDNA was amplified using KAPA HiFi HotStart ReadyMix (2×; KAPA Biosystems, Cat. No. KK2601) according to the manufacturer's protocol. The RNA probes were generated using the following   . Note the pale and dim OSP staining in cortical layer II/III in ipsilateral side compared with that in contralateral side. Similar reduction of OSP fluorescence intensity was seen in Kir4.1 cKO mice compared with its control. Scale bars: 50 µm. b The graph summary shows the percentage of OSP reduction in these two groups. The error bars represent s.e.m. n = 3 mice per group. Statistical significance was assessed as indicated using two-tailed Mann-Whitney test. c Electron micrographs demonstrate the presence of impaired axons with demyelination in ipsilateral cortex compared with its contralateral region after 30-45 min tMCAO mice at postnatal 8 weeks. In contralateral cortex of tMCAO, axons show normal myelin which exhibits dark, ring-shaped sheaths surrounding the axon, as indicated by arrowheads. Scale bar: 1 µm. d The magnified EM image shows a comparison of one myelin sheath surrounding axon in ipsilateral with its contralateral side after tMCAO. Scale bar: 1 µm. Bar graph and box plots represent average myelinated axon numbers, myelin sheath thickness, and G-ratio between ipsilateral cortex after tMCAO with its contralateral side. The data were normally distributed and statistical significance was assessed using two-tailed unpaired t-test, P-values are indicated, n represents analyzed axons from four mice. e Electron micrographs show the presence of impaired axons with demyelination in Kir4.1 cKO mice at postnatal 4 weeks compared with control mice. Scale bar: 1 µm. f The magnified EM image shows a comparison of one myelin sheath surrounding axon in Kir4.1 cKO mouse with its control. Scale bar: 1 µm. Bar graph and box plots represent average myelinated axon numbers, myelin sheath thickness, and G-ratio between Kir4.1-deficient mice with their control. The data were normally distributed and statistical significance was assessed using two-tailed unpaired t-test, P-values are indicated, n represents analyzed axons from three mice per group primers: NG2, Forward primer: GTTGGGATGCTTGCTGGTAT; Reverse primer: TGAAAGCTGCAGAAGCAGAA; Kir4.1, Forward primer: CTGCCCCGCGATT TATCAGA; Reverse primer: CATTCTCACATTGCTCCGGC. GAPDH, Forward primer: GGCAAATTCAACGGCACAGT; Reverse primer: TAGGGCCTCTCTT GCTCAGT.
For RNA-seq transcriptome experiment, PDGFRαCreER; mGFP mouse was anesthetized by isoflurane and the brain was removed. The fresh brain tissue was cut into small pieces and the minced tissue was incubated in 15 unit mL −1 papain at 31°C for 45 min. The digestion was stopped by protease inhibitor solution (Ovomucoid) 41 . After which, the tissue was immediately triturated and the isolated cells were seeded on coverslips. Single GFP-labeled NG2-glia was selected and aspirated into a glass pipette. The total RNA of NG2-glia in lysis buffer was converted to cDNA using the Smart-seq2 protocol and the cDNA was preamplified as described previously 40,42 . Illumina libraries were prepared using the commercially Sample Preparation kit (Nextera XT DNA Library Prep Kit) according to the manufacturer's instructions. The barcoded single-cell Illumina libraries of each experiment were pooled and sequenced for 2 × 75-base Paired-End reads on Illumina NextSeq500 sequencing system at the Sequencing Core of Shanghai Institute of Immunology, Shanghai Jiao Tong University School of Medicine. Sequencing reads were inspected by Fastqc 0.11.3 to check the reads quality and then aligned to the GRCm38/mm10 assembly of the mouse genome using Tophat 2.1.0 with the default options. FPKM (fragments per kilobase of exon per million fragments) values of each gene were obtained by Cufflinks 2.2.1 using genome annotation from UCSC (University of California, Santa Cruz). The GTF (gene transfer format) file was modified to update the genes encoding all inwardly rectifying potassium channel family members to the latest version archived in NCBI. To compare the expression level across different samples, FPKM values were transformed into TPM (transcripts per million) values after exclusion of microRNAs, small nucleolar RNAs, and rRNAs as previously reported 43 . The TPM values of all inwardly rectifying K + channel family members (Supplementary Table 1) were plotted as heat map as shown in Fig. 1c.
Electron microscopy. The procedure was conducted as described before with slight changes 44 . In brief, sections were rinsed in phosphate buffer and immersed in a solution of 1% osmium tetroxide in phosphate buffer for 1 h, then rinsed in phosphate buffer and gradually dehydrated on a series of ethanol from 30% to 70%. After that, the sections were stained with a solution of 1% uranyl acetate in 70% ethanol for 1 h and further dehydrated in ethanol. After dehydration was completed the sections were cleared in propylene oxide and infiltrated with Epon resin overnight at room temperature. The following day the sections were flat-embedded in new Epon resin and allowed to polymerize in an oven at 60°C for 72 h. Ultrathin sections (90 nm thick) were obtained using a Leica EM UC6 ultramicrotome (Leica Microsystems, Wetzlar, Germany), observed and photographed using a Hitachi TEM model H-7650 (Hitachi, Japan) equipped with an AMT digital camera (Danvers, MA). Selection of regions of interest was performed on a Nikon Eclipse 50i light microscope, carefully identifying anatomical regions and re-dissecting these regions for ultramicrotomy. The G-ratio for single labeled axons (longitudinally or transversally cut) was calculated from calibrated electron microscopy images as the diameter of the axon divided by the total diameter of the axon including the myelin sheath using ImageJ.
Data analysis. All statistical tests were run in GraphPad InStat 3. The graphs were created in Origin 8 and assembled in CorelDraw 12. Data are presented as mean ± s.e.m. For each set of data to be compared, we determined in GraphPad Instat whether the data were normally distributed or not. If they were normally distributed we used parametric tests, as listed in the text. If the data were not normally distributed we used non-parametric tests, as indicated in the text. Paired and unpaired Student's two-tailed t-tests (as appropriate and as indicated in the text) and two-tailed Mann-Whitney tests were used for most statistical analyses. For electrophysiological experiments, n values represent the number of recorded cells. For all biochemistry and immunohistochemistry experiments, n values represent the number of mice. Statistical significance was set at *P < 0.05, **P < 0.01, ***P < 0.0001, n.s., not significant.
Chemicals and drug application. All chemicals for electrophysiology were purchased from Sigma-Aldrich. The potassium blocker barium chloride was dissolved in double-distilled water at 50 mM and stored in aliquots at −20°C. The blocker was bath applied at 100 µM working concentration.
Data availability. The authors declare that all data supporting the findings of this study are available within the article and its supplementary information files. All relevant data not present within the manuscript or supplementary files are available from the corresponding author upon reasonable request. RNA-sequencing data are deposited at the Sequence Read Archive with SRA accession number SRP146737.