Animals

All animal procedures were approved and performed following the ethical guidelines set by the University of Virginia Institutional Animal Care and Use Committee (IACUC). Mice were housed in groups of five in a facility in a 12-h light–dark cycle with controlled temperature (21 ± 1.5 °C) and humidity (50 ± 10%) and had access to food and water ad libitum. Aldhe1l1-eGFP (GENSAT project) mice expressing eGFP under astrocyte-specific promotor AldheGFP were generated as described previously24 and housed and bred according to IACUC guidelines. We received C57BL/6N-Acantm1c(EUCOMM)Hmgu/H (European Mouse Mutant Archive stock EM:10224) from EUCOMM (UK Research & Innovation, Mary Lyon Center). The heterozygous mice were bred together to generate Acanfl/fl mice. To generate brain-wide developmental PNN KO, Acanfl/fl mice were bred with Nestin-Cre (B6.Cg-Tg(Nes-cre)1Kln/J, strain 003771-JAX) and confirmed by genotyping. PvTdTomato mice (C57BL/6-Tg (Pvalb-tdTomato) 15Gfng/J, strain 027395-JAX) and 5xFAD model mice of Alzheimer’s disease (B6SJL-Tg (APPSwFlLon,PSEN1*M146L*L286V) 6799Vas/Mmjax, strain 034840-JAX), FVB/NJ (strain 001800-JAX) and C57BL/6J (strain 000664-JAX) were purchased from The Jackson Laboratory and subsequently bred in an animal facility to generate experimental mice. We used adult 7–15-week-old mice of both sexes unless stated otherwise. All mice were genotyped to confirm the transgene expression or knockout before experimental use.

Intracranial surgeries and injectionsChABC injection

Chondroitinase ABC from Proteus vulgaris (C3667-10UN, Sigma-Aldrich) was dissolved in sterilized PBS (50 mU µl−1); subsequently, 2 µl solution was injected unilaterally at an infusion rate of 200 nl min−1. Mice were anesthetized with 2–5% isoflurane and fixed to a stereotaxic apparatus (Leica Angleone stereotaxic model, 39464710) followed by a midline scalp incision and a 0.5-mm burr hole 2.0 mm lateral and 1.0 mm ventral to bregma and infused ChABC at ~2.0-mm deep from the cortical surface using a 10-μl syringe (World Precision Instruments, SGE010RNS). Sham control mice were injected with sterile PBS with an identical procedure. Mice were dosed with buprenorphine/Rimadyl and allowed to recover on a heating pad until mobile and were monitored daily for up to 5 days after surgery. Body weight was measured for 3 consecutive days after surgery and all mice were perfused on day 6 after injection. Similar to previous studies32, we also found that ChABC injection homogeneously digests PNNs; however, faint and granular remains of the digested PNNs could still be visualized (Extended Data Fig. 5a,b) on intensity-enhanced images to identify previously intact PNNs.

AAV injection and Acan knockout

To knock out PNN in adult mice brains, we injected pENN.AAV.hSyn.HI.eGFP-Cre.WPRE.SV40 (Addgene, 105540-AAV9) in 7–8-week-old Acanfl/fl mice. In brief, AAV9 (2.7 × 1013 vg per ml) was diluted in PBS to achieve 1 × 1013 vg per ml concentration and 1.5 µl was injected in each hemisphere (from bregma: 0.5 mm posterior, 2.0 mm lateral and 1.0 mm ventral) with 200 nl min−1 infusion rate as described above. Mice were transcardially perfused after 8–10 weeks of AAV9 injections to perform IHC.

AAV injection and iGluSnFR expression in astrocytes

Male and female PvTdTomato mice (6–12-week-old) were intracranially injected (2.0 mm lateral and 0.25 mm caudal from bregma on both hemispheres) with iGluSnFR AAV (pAAV.GFAP.SF-iGluSnFR.A184S AAV1, Addgene 106192-AAV1). Then, 1.20 µl of iGluSnFR AAV (1.6 × 1013 genome copies per ml) was injected at a rate of 0.12–0.15 µl min−1 using a 10-µl syringe (Hamilton, CAL800000 1701N). Mice were dosed with buprenorphine/Rimadyl and allowed to recover on a heating pad until mobile and were monitored daily for up to 5 days post-surgery. All mice receiving AAV injections were used for experiments within 3–5 weeks of AAV injection.

Acute slice electrophysiology

Whole-cell patch-clamp recordings were obtained from astrocytes in situ acute brain slices as described previously21. In brief, mice underwent cervical dislocation followed by a quick decapitation and dissection to remove brains and were kept in an ice-cold ACSF (135 mM NMDG, 1.5 mM KCl, 1.5 mM KH2PO4, 23 mM choline bicarbonate, 25 mM d-glucose, 0.5 mM CaCl2 and 3.5 mM MgSO4, pH 7.35, 310 ± 5 mOsm) (all from Sigma-Aldrich) saturated with carbogen (95% O2 + 5% CO2). We prepared 300-μm thick coronal slices using Leica VT 1000P or 1200S tissue slicers. Slices were transferred into a custom-made recovery chamber filled with ACSF (125 mM NaCl, 3 mM KCl, 1.25 mM NaH2PO4, 25 mM NaHCO3, 2 mM CaCl2, 1.3 mM MgSO4 and 25 mM glucose, pH 7.35, 310 ± 5 mOsm) constantly bubbled with carbogen (95% CO2 + 5% O2) to recover at 32 °C for 1 h. Subsequently, slices were transferred to room temperature conditions until used for recordings. Individual slices were transferred to a recording chamber that was continuously superfused with ACSF at a flow rate of 2 ml min−1. GFP-positive astrocytes in Aldh1l1-eGFP mice cortical slices were visualized using an upright microscope (Leica DMLFSA) with ×5 and ×40 water-immersion lens and epifluorescence and infrared illuminations to identify eGFP-expressing astrocytes.

Whole-cell voltage-clamp and current-clamp recordings were conducted using an Axopatch 200B amplifier (Molecular Devices) with an Axon Digidata 1550A interface (molecular devices). Patch pipettes of 7–10 MΩ open-tip resistance were created from standard borosilicate capillaries (WPI, 4IN THINWALL Gl 1.5 OD/1.12 ID) using HEKA PIP 6 (HEKA) or PMP-102 (Warner Instruments) programmable pipette pullers. We filled patch pipettes with an intracellular solution containing 134 mM potassium gluconate, 1 mM KCl, 10 mM 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid (HEPES), 2 mM adenosine 5′-triphosphate magnesium salt (Mg-ATP), 0.2 mM guanosine 5′-triphosphate sodium salt (Na-GTP) and 0.5 mM ethylene glycol tetraacetic acid (EGTA) (pH 7.4, 290–295 mOsm). MM-225 micromanipulator (Sutter Instrument Co.) was used to visually guide the patch pipette to the cell. After making a tight seal of >5 GΩ resistance, brief suction was applied to achieve the whole-cell mode and cells were immediately clamped at −80 mV. The membrane capacitance (Cm) and series resistance were not compensated. Data were acquired using Clampex 10.4 software and Axon Digidata 1550A interface (Molecular Devices), filtered at 5 kHz, digitized at 10–20 kHz and analyzed using Clampfit 10.6 or Clampfit 11.2 software (Molecular Devices). Unless otherwise stated, throughout all the recordings carbogen-bubbled ACSF was continuously superfused (2 ml min−1) and the bath temperature inside the reordering chamber was maintained at 32–33 °C using an inline feedback heating system (TC 324B, Warner Instruments).

PNN degradation in ex situ brain slices

ChABC from Proteus vulgaris (C3667, Sigma-Aldrich) was reconstituted in 0.01% bovine serum albumin aqueous solution according to the manufacturer’s instruction to make a 1 U per 40 μl stock solution. Aliquots of 1 U were prepared and stored at −20 °C until used. After slice recovery, slices were treated with ChABC and subsequent recordings were made as previously described14,15. In brief, after recovery, 2–3 cortical half slices were incubated in 3 ml ChABC solution (0.5 U ChABC per ml in ACSF) in an incubation chamber continuously supplied with carbogen at 33 °C for 45 min. Next, slices were rinsed twice and incubated in ACSF until used for electrophysiological recordings. These parameters of PNNs digestion by ChABC (enzyme concentration of 0.5 U ml−1, incubation time of 45 min and incubation temperature of 33 °C) reliably degraded PNNs (Extended Data Fig. 7a) as described previously14,15. For controls, previously separated contralateral halves of the ChABC-treated slices were incubated in 3 ml ACSF without ChABC and subsequently, both ChABC-treated and non-treated slices were kept in ACSF together until used for the recordings.

Measurement of intrinsic biophysical properties of astrocytes

The resting membrane potential (Vm) was measured by setting I = 0 mode immediately after achieving the whole-cell configuration. The Cm was measured directly from the amplifier by adjusting capacitance and monitoring the capacitive transients as described previously21. To calculate the input resistance (Rin) of astrocytes, we calculated the steady-stated membrane voltage deflection (ΔV) on injecting 15 hyperpolarization current steps (−100 pA each for 1,000 ms). The ratio (ΔV/I) of steady-state change in the membrane voltage (ΔV) and the corresponding injected current (I) was computed as Rin. The I–V curve of astrocytes was computed in both the current clamp (31 steps, −100 pA to +500 pA, step size 20 pA, step duration 1,100 ms) and voltage clamp (25 sweeps, −180 mV to +60 mV, step size 10 mV) modes (Extended Data Fig. 7f–i) followed by plotting the voltage/current responses. Astrocytes with nonlinear I–V responses were not continued for recordings and analysis.

Measurement of astrocytic currentsSynaptically evoked Glu uptake current

We recorded synaptically evoked currents from cortical astrocytes according to the previously published studies with some modifications23. In brief, we placed a concentric bipolar electrode (FHC, CBABD75) in L5–6 of the cortical slices and patched astrocytes in L3–4 of the somatosensory area (Fig. 6b). The stimulation protocol consists of initial 10-µA and 20-µA pulses followed by a 20-µA increment in each subsequent pulse capping at 200 µA (pulse duration 200 µs). All recordings were performed in the presence of 20 µM bicuculline, 50 μM d-AP5, 20 μM CNQX and 100 μM BaCl2. In the initial few recordings, we confirm that the recorded current is glutamate by observing a near-complete blockade of evoked current upon 100 μM TBOA and 300 μM DHK application (Extended Data Fig. 7j). The remaining current was abolished by superfusing 0.5 μM TTX, confirming it as a neuronal-evoked Glu current (Extended Data Fig. 7j). Each stimulation pulse was repeated five times (sweeps) and a minimum of two sweeps were averaged to compute the peak current and charge transfer after excluding the sweeps with baseline fluctuation or noise.

Depolarization evoked potassium uptake current

To record depolarization evoked astrocytic potassium uptake current, we positioned the stimulator and patch pipette as described above and incubated slices in a mixture of 20 µM bicuculline, 50 μM d-AP5, 20 μM CNQX, 100 μM TBOA and 300 μM DHK. The stimulation protocol consisted of initial 0.1-mA and 0.2-mA pulses followed by 0.5 mA and five subsequent pulses with 0.5-mA increments capping at 3 mA (pulse duration 200 µs). Each stimulation pulse was repeated three times (sweeps) and a minimum of two sweeps were averaged to compute the peak current and charge transfer after excluding the sweeps with baseline fluctuation or noise.

Astrocytic uptake of exogenously applied Glu

To measure the Glu uptake capacity of astrocytes we adopted the exogenous Glu puffing method as described previously with minor modifications21,24. In brief, we constantly perfused slices with ACSF containing 500 nM TTX, 20 μM bicuculline, 100 μM CdCl2, 50 μM d-AP5 and 50 μM CNQX and 100 μM BaCl2. After patching an astrocyte, a 500-ms puff (2 psi pressure using a Pico-liter Injector PLI-10 from Warner Instruments) of 200 μM glutamate solution (120 mM NaCl, 3.5 mM KCl, 25 mM HEPES, 10 mM glucose and 200 µM Glu) was applied from a distance of ~100 μm by a 5–8 MΩ open-tip resistance glass pipette. In several random recordings, we applied a mixture of 100 μM TBOA and 300 μM DHK to confirm that the recorded current was glutamate (Fig. 6i). We recorded five sweeps and averaged a minimum of three sweeps to generate a result sweep that was utilized to compute the data. The sweeps with fluctuating baseline and noise were excluded from the analysis. The averaged trace of uptake current was analyzed using Clampfit 10.6 or Clampfit 11.2 program to generate the below-described measurements. The peak current was calculated by subtracting the baseline from the peak response. The total charge transfer was computed by calculating the total areas under the curve of Glu uptake current response. Decay time and decay slope were calculated from the decaying phase (100% to 37% of the peak) of the uptake current.

Two-photon microscopy imaging of astrocytic glutamate uptake using iGluSnFR

For live imaging of iGluSnFR in two-photon microscopy, recovered slices were placed in a recording chamber with continuous superfusion of carbogen-saturated ACSF at 2–3 ml min−1 with bath temperature maintained at 32–33 °C with an inline heater. Images were captured using a four-channel Olympus Dual-beam FVMPE-RS multiphoton microscope (Olympus/Evident) equipped with two high-sensitivity cooled GaAsP detectors, two multialkali photomultiplier detectors, a high-resolution Galvo and a fast resonant scanner, an XLPLN ×25/1.05 NA water-immersion objective (Olympus/Evident) and an InSight X3 ultrafast pulsed laser (Spectra Physics) with tuning range of 680–1,300 nm. For all experiments, the InSight laser was tuned to 870 nm for simultaneous excitation of both iGluSnFR and TdTomato. Olympus Fluoview software (Olympus/Evident) was used for all data acquisition and in combination with CellSens (Olympus/Evident) and Clampfit (Molecular Devices) software programs for subsequent analysis.

Slices were placed with the ventral side facing the field stimulating electrode and observed through the epifluorescence microscope to confirm TdTomato expression of PV cells (555 nm) and iGluSnFR fluorescence (488 nm). Areas of recording were determined by the presence of a PvTdTomato+ PV neuron with astrocyte presence in the immediate periphery/boundary of the cell. Single-frame images of the cells were taken using the high-resolution Galvo scanner at 1,024 × 1,024 pixels with a zoom of ×6. Imaging of realtime stimulation and subsequent changes of iGluSnFR expression was performed using the fast resonant scanner in a bidirectional scanning mode, imaging a region of 512 × 127 pixels with a zoom of ×6 resulting in an acquisition rate of approximately 40 Hz.

Synaptic release of Glu was achieved using a constant current isolated stimulator model DS3 (Digitimer) to emit a single pulse of current from the paired field stimulation electrode. Placement of the electrode consisted of locating the strongest expression of fluorescence in cortical L3–4 of the SSC region and tracing the neurites down to the corresponding cortical L5–6. Upon identification of a position with good fluorescence in both L3–4 and L5–6, the electrode was lowered into the bath and placed at the L5–6 location with very slight embedding into the surface of the slice. Following the placement of the electrode, the field of view was moved back up to the corresponding L3–4 of the SSC for visualization and recording of the fluorescent signal. A waveform generator (Olympus/Evident) was used to trigger the stimulation at an identical timepoint for each recording. To reliably detect the signal, we injected longer pulses of field stimuli than in previously described electrophysiology experiments, using stimulations of 10 ms of 500 μA.

All recordings consist of a brief baseline period followed by stimulation and finally a brief baseline recovery period. All recordings were performed first in ACSF followed by perfusing TBOA and DHK for 5–7 min and a repeated recording with TBOA and DHK in the bath. To minimize photobleaching and subsequent errors, we shortened the recording period and corrected the baseline using automated functions in the Clampfit program. We computed the peak intensities using the Clampfit program followed by subtracting the peak intensity before (ACSF) from that obtained after TBOA + DHK incubation (TBOA + DHK) to obtain the net change ((TBOA + DHK) − ACSF). To account for the baseline variability in fluorescence intensity, we used the peak intensity before TBOA + DHK (ACSF) as 100% and normalized the peak response after TBOA + DHK incubation (TBOA + DHK) and generated % net change (normalized to 100). We computed the % net change from control and ChABC-digested groups and performed statistical analysis.

Seizure induction and EEG recordingsTMEV-induced seizures

We used Daniel’s strain of TMEV to induce seizures in mice. TMEV was provided by the laboratories of K. S. Wilcox and R. S. Fujinami from the University of Utah. The titer of the stock used was 2 × 107 plaque-forming units (p.f.u.) per ml. Anesthetized mice (3% isoflurane) were injected with 20 µl of either PBS or TMEV solution (2 × 105 p.f.u.) intracortically by inserting a 28-gauge needle perpendicular to the skull surface, slightly medial to the equidistant point on the imaginary line connecting the right eye and the right ear. We used a sterilized syringe containing a plastic sleeve on the needle to expose only 2.5 mm of the needle from the tip to restrict the injection in a cortical region. The needle was slowly retracted after 1 min of injection followed by disinfecting the injection site. Mice started behaving normally after 5–10 min of the injection.

To assess the handling-induced acute behavioral seizures between 2–8 days post-injection, we briefly agitated the cage by shaking and observed mice behavior for about 5 min. Seizures occurred within 1 min of handling and the seizure score was assessed using a modified Racine scale with stage 1 (mouth and facial movements); stage 2 (head nodding); stage 3 (forelimb clonus); stage 4 (forelimb clonus and rearing); stage 5 (forelimb clonus, rearing and falling); and stage 6 (intense running, jumping, repeated falling and severe clonus18).

PNN disruption and EEG electrode implantation surgery

A mixture of ChABC (C3667-10UN, Millipore Sigma) and hyaluronidase (H1136-1AMP, Millipore Sigma) enzymes was injected intracortically to degrade ECM in mice (FVB/N, The Jackson Laboratory; 8 weeks old). Stock solutions of ChABC (0.05 U µl−1) and hyaluronidase (2 U µl−1) were prepared by reconstituting lyophilized enzyme powder in sterile saline. The stocks were aliquoted in a single-use amounts (10 µl) to avoid thaw–freeze cycles and stored at −80 °C.

Mice were anesthetized using 3% isoflurane, provided analgesia (0.1 mg kg−1 buprenorphine and 5 mg kg−1 carprofen intraperitoneally) and heads were affixed into a stereotaxic instrument (David Kopf Instruments). The hair over the skull area was removed using a hair-removal cream, the surgical area was disinfected using iodine and 70% alcohol and the skull was exposed. Mice were continuously anesthetized using nasal tubing supplying 1–2% isoflurane throughout the surgical procedure. The aliquots of ChABC and hyaluronidase were thawed on ice just before use. An injection mixture of ChABC, hyaluronidase and saline was prepared in a ratio of 2:2:1 (10 µl ChABC + 10 µl hyaluronidase + 5 µl saline) sufficient for two mice. The enzyme preparations were kept on ice throughout the procedure. The control group of mice received saline injections. Two injection holes were drilled in the skull (−1.0 mm ML (mediolateral) from bregma, 1.0 mm or −2.0 mm AP (anteroposterior) from bregma). The enzyme mixture was then injected stereotactically into the cortex by inserting the needle into the brain at 45° angle to target the injections at −2.0 mm ML from bregma, 1.0 mm or −2.0 mm AP from bregma, and −0.5 DV (dorsoventral) from the brain surface. Neuros syringe (Model 1701, 65460-06, Hamilton) and 33-gauge needle (65461-02, Hamilton) were used to inject 4.5 µl solution per injection site at 0.6 µl min−1 injection speed (Quintessential Stereotaxic Injector, 53311, Stoelting). The needle was checked for any blockage by pumping a small amount of fluid just before inserting it into the brain. The needle was kept undisturbed for 2 min after injection and slowly retracted to prevent leakage of any fluid. The holes in the skull were filled with bone wax. With this procedure, a single mouse received a total of 0.18 U of ChABC and 7.2 U hyaluronidase in the left cortex.

To implant electrodes for the EEG recordings, a total of six holes (three for anchor screws and three for electrodes) were drilled in the skull carefully without causing bleeding. Two electrodes of a three-channel electrode set up (MS333/8-A, P1 Technologies) were implanted in the cortex bilaterally using stereotaxic coordinates of ±2.0 mm lateral and 1.5 mm posterior from bregma and about 0.5 mm ventral from the brain surface. The ground electrode was implanted over the brain surface near the cerebellum (−1.0 mm lateral and 5.0 mm posterior from bregma). Three anchor screws were carefully inserted into the skull without damaging the brain surface; the first one anterior to the bregma in the right hemisphere, the second one over the left parietal cortex and the third one posterior to the lambda in the right hemisphere. The electrodes and screws were secured in position using dental cement and the skin incision was closed using tissue glue. All surgically operated mice were treated humanely and provided with postoperative care as per the National Institutes of Health guidelines and the IACUC protocol.

Video EEG acquisition and seizure analysis

The vEEG was initiated after 3 h of surgical procedure and acquired continuously for 7 days. Mice were connected to an EEG100C differential amplifier (BIOPAC Systems) using a lightweight three-channel cable with a three-channel rotating commutator (P1 Technologies). The MP160 data acquisition system and AcqKnowledge 5.0 software (BIOPAC Systems) were used to record electroencephalograms. M1065-L network camera (Axis Communications) and media Recorder 4.0 software (Noldus Information Technology) were used to record the behavior of each mouse. All the cables and electrical components were sufficiently shielded to minimize electrical noise. Video and EEG recordings were automatically synchronized using Observer XT 14.1 software (Noldus Information Technology). EEG signals were bandpass filtered between 0.5 and 100 Hz, amplified and digited at a sampling frequency of 500 Hz. Mice had access to food and water conveniently during the entire vEEG recording.

The EEG and video recordings were reviewed manually by an experimenter blinded to the treatment group of mice. Electrographic seizures were defined as rhythmic spikes or sharp-wave discharges with amplitudes at least twice the average amplitude of baseline, frequency at least 2 Hz and duration at least 5 s. Behavioral seizures were also identified by verifying postictal suppression of the baseline EEG activity, which typically occurs after a seizure but is not accompanied by electrographic artifacts associated with mouse behavior other than seizures. Seizure duration and seizure latency (time to occurrence of first seizure after intracortical injections) were calculated from the electrographic seizure data. Seizure severity was graded using a Racine scale as follows: stage 1 (mouth and facial movements); stage 2 (head nodding); stage 3 (forelimb clonus); stage 4 (forelimb clonus and rearing); and stage 5 (forelimb clonus, rearing and falling). Subtle behavioral seizures without forelimb clonus were assigned a seizure score <3. At the end of the experiment, mice were perfused transcardially with saline to remove blood and with 4% paraformaldehyde to fix the brains for IHC analysis to assess the degradation of ECM.

Gel electrophoresis and western blot

The levels of Kir4.1 and GLT1 in the acute slices after ChABC treatment were measured by western blot analysis with some modifications as described previously24,25. In brief, we prepared acute cortical slices, followed by recovery and ChABC-mediated PNN digestion as described above. Cortical tissue from brain slices was isolated and flash-frozen by briefly dropping tubes containing tissue into liquid nitrogen and storing at −80 °C until further processing. We homogenized the tissue using a rotor-stator homogenizer in a lysis buffer (50 mM Tris-HCl, pH 8.00, 150 mM NaCl, 1% Triton X-100, 0.5% sodium deoxycholate, 0.1% SDS, protease inhibitors (P8340, Sigma) and phosphatase inhibitors (P0044, Sigma); 10 µl lysis buffer per mg of tissue), followed by collecting the supernatant and centrifugation (15,000g, 20 min, 4 °C). We used bicinchoninic acid (BCA) protein assay (Pierce BCA Protein Assay kit, 23225, Thermo Fisher Scientific) to determine the total protein concentration in the supernatant. Next, we denatured a 10-µg total protein sample at 50 °C for 10 min followed by performing electrophoresis using polyacrylamide gel (4–15% Tris–glycine extended polyacrylamide gel, 567-1085, Bio-Rad). Proteins were transferred from gel to a PVDF membrane, followed by blocking in Tris-buffered saline-based Odyssey blocking buffer (LI-COR) for 1 h at room temperature. Next, we incubated the membrane with rabbit anti-Kir4.1 (1:2,000 dilution, APC-035, Alomone Labs) or guinea pig anti-GLT1 (1:1,500 dilution, AB1783 Millipore) and mouse-anti-β-actin antibody (1:3,000 dilution, MA1-140, Invitrogen) overnight at 4 °C followed by washing and incubation with secondary antibodies (IRDye 800CW donkey anti-guinea pig/rabbit IgG, 0.2 µg ml−1, 925-32212 and IRDye 680RD donkey anti-mouse IgG, 0.2 µg ml−1, 925-68073; LI-COR) at 1:20,000 dilution for 2 h at room temperature. Secondary antibodies were washed and membranes were imaged using an Odyssey imaging system (LI-COR). Densitometric analysis of protein levels was performed using Image Studio software (LI-COR).

Immunohistochemistry

Mice were injected with a mixture of ketamine and xylazine (100 mg kg−1 and 10 mg kg−1, respectively) and subsequently perfused transcardially with PBS followed by 4% PFA. We dissected out the brains and stored them overnight in 4% PFA at 4 °C followed by storing them in PBS at 4 °C until sectioning was performed. We cut 50-μm-thick floating sections using a 5,100 mz vibratome from Campden Instruments or Pelco EasiSlicer from Ted Pella. The sections were either used for IHC immediately or stored at −20 °C in a custom-made storage medium (10% (v/v) 0.2 mM phosphate buffer, 30% (v/v) glycerol, 30% (v/v) ethylene glycol in deionized water, pH 7.2–7.4) for future uses. To minimize procedure-associated variations, we stained duplicate sections from five to seven mice of each experimental group in a single batch. In brief, sections were retrieved from −20 °C storage, rinsed three times with PBS and permeabilized and blocked by incubating them in blocking buffer (0.5% Triton X-100 and 10% goat serum in PBS) for 2 h at room temperature in a 24-well plate. Sections were incubated for 18–24 h at 4 °C with appropriate primary antibodies or biotinylated WFA (B-1355, Vector Laboratories) in diluted blocking buffer (1:3 of blocking buffer and PBS). Following this, we incubated sections with appropriate secondary antibodies and Alexa Fluor 555-conjugated streptavidin (S32355, Thermo Fisher Scientific, 1:500 dilution) in diluted blocking buffer overnight at 4 °C in dark. Further, the sections were rinsed with PBS and were mounted on glass slides (Fisherfinest 25 × 25 × 1, 12-544-2) covered with cover glass and the edges of the slides were sealed with nail polish. Amyloid plaques in 5xFAD+ mice sections were stained by Amylo-Glo (TR-300-AG, Biosensis) following the manufacturer’s instructions. All antibodies used in the study were validated either by vendors and/or many published studies. The primary and secondary antibodies used are given in Supplementary Table 1.

Confocal imaging and analysis

Representative images and data in Figs. 1a–f, 2a–e and 4b–r and Extended Data Figs. 1 and 3) were acquired using Nikon A1 confocal microscope, and quantification was performed by associated NIS-Elements AR analysis program. Images and data in the remaining figures were acquired using an Olympus FV 3000 confocal microscope and images were analyzed using ImageJ. We utilized several different objective lenses, including ×10 (air), ×20 (air), ×40 (oil), ×60 (oil) or ×100 (oil) with a range of optical zoom based on the experimental requirement. Images were acquired as 12 bits and acquisition settings were minimally adjusted to accommodate a few unsaturated and saturated pixels.

To acquire high-magnification images for PNN hole analysis, we excluded the topmost ~2-µm area from the surface due to the occasional uneven tissue surface and restricted our imaging to the top 3–10-µm depth (~8-µm-thick tissue block). Within the above-defined ~8-µm tissue block, we selected the optical plane containing the largest perimeter of a PNN/PV neuron. We used either a ×40 oil immersion objective lens (Plan Fluor ×40 Oil DIC H N2, 1.3 NA, Scanner zoom 4 or 5, 0.2 µm optical section) or ×100 oil immersion (UPlanApo 100XOHR 1.5 NA, 3 or 4 zoom, 0.2-µm optical section, 0.031 µm per pixel) objective lens to take 1,024 × 1,024-sized images. With a range of 405–647 nm light wavelength, the resolution limit of the above objective lenses is 250–310 nm. With the above settings, the lateral and axial resolution exceeded the Nyquist to reliably digitize the optical signal. We adjusted the acquisition settings (laser power, PMT gain and offset) to accommodate the full range of signal (reflected by a few under-saturated and few saturated pixels in the image). Once set, we minimally adjusted the acquisition settings.

PNN disruption analysis

The spread of PNN disruption by ChABC injection in mice brains was quantified from whole coronal section images (Fig. 4b,c). We drew uniform-sized regions of interest (ROIs) (0.4 × 0.4 mm2) adjacent to each other starting from the ChABC incision site toward the lateral side of the coronal plane. The mean fluorescence intensity was computed. All analyzed images/ROIs at similar distances from the incision site were tabulated to plot the mean and s.d. of the fluorescence intensity. To assess the PNN disruption on PV cells after AAV-mediated Acan KO, 5xFAD model of Alzheimer’s disease and TMEV model of seizure, we selected a 0.8-µm perimeter area of cell soma and binarized the WFA signal using an automated thresholding method (OTSU) in ImageJ and computed WFA intensity and pericellular WFA area (Fig. 5).

Analysis of PNN holes

We assessed the PNN holes for the presence of astrocytic processes (Fig. 1) and synaptic contacts (Fig. 2) and their fate after ChABC treatment (Extended Data Fig. 5) using the PNN line intensity profile method with slight modifications as described previously14,15. In brief, we acquired high-magnification (×200 or higher) images of individual PNNs at their maximum perimeter plane (Supplementary Fig. 1). Subsequently, we drew a polyline on PNN (WFA) along the entire periphery of the cell and generated an intensity profile consisting of high-intensity peaks (Supplementary Fig. 1b, magenta plot) and low-intensity drops. We set a threshold of WFA intensity (ranging from 40–66% of the unsaturated peak WFA intensity) that covered the maximum number of drops as PNN holes (Supplementary Fig. 1b). We observed that the size of a hole in the line profile may range from several microns to as small as 0.5 µm (Supplementary Fig. 1c, gray bars) depending on whether the image was captured at the center or the edge of a hole. To overcome this subjectivity, we set a cutoff of 0.5 µm (based on the 250–310 nm resolution limit of objective lenses) as the minimum width of a hole to consider for analysis. We moved a 0.5-µm bar through the entire line profile and marked the holes with arrows (Supplementary Fig. 1c, black arrows) for counting purposes. Thus, a minimum 0.5-µm wide WFA intensity drops under the threshold line (Supplementary Fig. 1c–e) were considered as PNN holes.

The presence of a specific fluorophore peak in the PNN hole was determined by the presence of a clearly distinguishable peak within the two consecutive peaks of WFA (Supplementary Fig. 1c–e, gray bars). Subsequently, we provided unique identifying marks (Supplementary Fig. 1c, black downward arrows) to each PNN hole and computed the presence or absence of astrocytic/synaptic components within it. To quantify the degree of perforations in PNNs after ChABC disruption, we counted the number of PNN holes as described above and normalized it to the cell perimeter.

PNN disruption and Kir4.1 and GLT1 expressions analysis in fixed acute brain slices

To assess whether ChABC-mediated PNN digestion in acute brain slices alters the expressions of Kir4.1 and GLT1 to influence the astrocytic potassium and glutamate uptake, we fixed acute brain slices after electrophysiological recordings and performed immunostaining using specific antibodies and WFA. The ×40 magnification images were acquired using Olympus FV 3000 and analyzed using ImageJ. The signal of the individual channel (Kir4.1/GLT1, WFA and AldheGFP) was binarized using an inbuilt thresholding function OTSU and the resulting total area was tabulated to plot the graphs.

Quantification of astrocytic coverage and synaptic puncta

To quantify the pericellular astrocytic coverage of PV/excitatory neurons, we acquired high-magnification images using either Nikon A1 (40 × 5 optical zoom oil immersion lens) or Olympus Fluoview FV 3000 (100 × 3 or 4, oil immersion objective lens) with a 0.2-µm optical plane thickness. After image acquisition, we generated a binary representation of the cell soma using inbuilt functions in ImageJ and Nikon elements programs. We defined a 0.8-µm wide perimeter from the cell surface as a pericellular area (based on the pericellular width covered by PNN). Subsequently, we binarized the individual channels (AldheGFP, S100B, GLT1 and Kir4.1) using inbuilt auto thresholding functions in Nikon elements or ImageJ (OTSU). Using Boolean operations, we computed the binary areas of different astrocytic markers confined to the cell perimeter defined above (Supplementary Fig. 2). We normalized the pericellular area with the perimeter of the same cell before pooling images from mice.

We added one more step of find maxima in ImageJ or an analogous function in Nikon AR analysis programs to quantify the overall and pericellular numerical densities of vGlut1 and vGAT puncta in the above protocol. A prominence setting of 500 (for vGlut1 puncta) or 2,000 (for vGAT puncta) was found appropriate to capture all puncta and was used for images. The total number of synapses in the entire image was used to plot the total vGlut1/vGAT puncta. We used Boolean operations to compute the total pericellular synapses and pericellular synapses with astrocytic processes in contact with them. The resultant absolute values were normalized to the perimeter of the individual cells and were used for data pooling or directly for plotting graphs.

Volumetric analysis in Imaris

The representative 3D rendering videos and 3D reconstruction images of the PNN and pericellular astrocytic processes in Figs. 1, 2, 3 and 4 were generated using Imaris v.9.90 (Oxford Instruments). In brief, we generated volumetric masks from the PV channel that were expanded by 0.8 –1.0 µm to accommodate pericellular PNN structures. These masks were then used to create new astrocytic and synaptic data channels that excluded structures outside of the pericellular domain. The enlarged PV channel volume was created using the surface creation tool with smoothing detail enabled and a surface grain size set to 0.103 µm. Background subtraction was also enabled with the diameter of the largest sphere set to 0.388 µm and manual thresholding set to a value of 200. Astrocytic and synaptic channel volumes were created using the surface creation tool with smoothing detail enabled and a surface grain size set to 0.103 µm. Background subtraction was also enabled with the diameter of the largest sphere set to 0.388 µm and manual thresholding set to 200.

Illustrations and cartoons were created with BioRender.com.

Statistics and reproducibility

Data in the bar diagrams are expressed as mean ± s.d. unless stated otherwise in the figure. Individual data points are represented by dots. Figure legends contain the essential details, including numerical values of mean, s.d., biological or technical replicates, statistical tests and corrections. The detailed statistical analysis data, including test statistics, P values, post hoc comparisons and corrections are summarized in Supplementary Table 2. The sample size was not predetermined but was based on published relevant studies. We ran appropriate normality tests and found that data distribution was sufficiently normal and variance within groups was sufficiently similar to be used for parametric tests. Therefore, experimental designs with two treatment groups were analyzed by two-tailed unpaired t-test unless stated otherwise in the figure legends. Welch’s correction was applied regardless of statistically different variances unless stated otherwise. Experimental designs with more than two groups were analyzed using one-way or two-way ANOVA followed by Tukey’s post hoc multiple comparison tests. Statistically significant differences between groups are shown in graphs as *P < 0.05, **P < 0.01, ***P < 0.001 and ****P < 0.0001. No data were excluded. Data analysis was performed using Microsoft Excel and Origin 2021b (OriginLab). For representative experiments, including Figs. 3b,c,j and 6a and Extended Data Figs. 1a,d,f,h, 2a, 5a,b and 9a, we conducted a minimum of three observations in three different mice. Data collection and analysis were performed blind to the conditions of the experiments for data in Figs. 3l–q, 5, 6q and 7l,m and Extended Data Figs. 4 and 6; for remaining data, explicit visual differences in experimental groups prevented us from performing blinded experimentation and analysis.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.