Neuronal gamma oscillations and activity‐dependent potassium transients remain regular after depletion of microglia in postnatal cortex tissue

Microglial cells (resident macrophages) feature rapid activation in CNS disease and can acquire multiple phenotypes exerting neuroprotection or neurotoxicity. The functional impact of surveying (“resting”) microglia on neural excitability and neurotransmission in physiology is widely unknown, however. We addressed this issue in male rat hippocampal slice cultures (in situ) by pharmacological microglial ablation within days and by characterizing neuronal gamma‐band oscillations (30–70 Hz) that are highly sensitive to neuromodulators and disturbances in ion and energy regulation. Gamma oscillations support action potential timing and synaptic plasticity, associate with higher brain functions like perception and memory, and require precise communication between excitatory pyramidal cells and inhibitory (GABAergic) interneurons. The slice cultures featured well‐preserved hippocampal cytoarchitecture and parvalbumin‐positive interneuron networks, microglia with ramified morphology, and low basal levels of IL‐6, TNF‐α, and nitric oxide (NO). Stimulation of slice cultures with the pro‐inflammatory cytokine IFN‐γ or bacterial LPS serving as positive controls for microglial reactivity induced MHC‐II expression and increased cytokine and NO release. Chronic exposure of slice cultures to liposome‐encapsulated clodronate reduced the microglial cell population by about 96%, whereas neuronal structures, astrocyte GFAP expression, and basal levels of cytokines and NO were unchanged. Notably, the properties of gamma oscillations reflecting frequency, number and synchronization of synapse activity were regular after microglial depletion. Also, electrical stimulus‐induced transients of the extracellular potassium concentration ([K+]o) reflecting cellular K+ efflux, clearance and buffering were unchanged. This suggests that nonreactive microglia are dispensable for neuronal homeostasis and neuromodulation underlying network signaling and rhythm generation in cortical tissue.

To sense danger signals and/or homeostatic imbalance within the brain parenchyma, microglia express a variety of receptors that recognize, for example, bacterial and viral components, modified endogenous ligands, neurotransmitters, cytokines, and chemokines (Deczkowska et al., 2018;Kettenmann, Kirchhoff, & Verkhratsky, 2013). This sensing is facilitated by highly dynamic cellular processes that monitor a certain territory and periodically contact neuronal synapses, including the enwrapping astrocytes, even in surveying ("resting") microglia under physiological conditions (Kettenmann et al., 2013;Sierra et al., 2010).
We characterized cholinergic neuronal gamma-band oscillations  and stimulus-induced transients of the extracellular K + concentration ([K + ] o ) in the presence and absence of microglia.

Significance
Microglia are resident immune cells that become reactive in most brain diseases, such as stroke, bacterial meningoencephalitis, and Alzheimer's disease. The role of nonreactive (surveying) microglia during normal brain information processing is widely unknown, however. We show that neuronal gamma oscillations (30-70 Hz) reflecting precise communication between different types of neurons as well as extracellular potassium ion dynamics reflecting the activities of neurons and astrocytes remain regular after pharmacological removal of microglia from postnatal cortex tissue. Our data support the biological concept that nonreactive microglia are dispensable for neuronal homeostasis, signaling, and rhythm generation in the healthy postnatal brain.

| Animals
Wistar rats were purchased from Charles River (Sulzfeld, Germany) and handled in accordance with the European directive 2010/63/EU and with the consent of the animal welfare officers at the University of Heidelberg (licenses, T56/11, T46/14, and T96/15). Experiments were performed and reported in accordance with the ARRIVE guidelines.

| Preparation and exposures of slice cultures
Organotypic hippocampal slice cultures were prepared as described (Kann et al., 2003;Papageorgiou et al., 2016). In brief, hippocampal slices (400 µm) were cut with a McIlwain tissue chopper (Mickle Laboratory Engineering Company Ltd., Guildford, UK) from 49 male rats at postnatal day 9 or 10 (p9-p10) under sterile conditions. Male rats were used to avoid the effects of sex-related variability in microglial and neuronal biology (Bordt, Ceasrine, & Bilbo, 2020). Slices with incomplete hippocampal structures were rejected. Three to five slices were maintained on Biopore™ membranes (Millicell standing inserts, Merck Millipore, Darmstadt, Germany), at the interface between serum-containing culture medium and humidified normal atmosphere enriched with 5% CO 2 (36.5°C) in an incubator (Heracell, Thermoscientific, Dreieich, Germany). The culture medium consisted of 50% minimal essential medium, 25% Hank's balanced salt solution (Sigma-Aldrich, Taufkirchen, Germany), 25% heat-inactivated horse serum (Life Technologies, Darmstadt, Germany), and 2 mM L-glutamine (Life Technologies) at pH 7.3 titrated with Trisbase. The culture medium (1 ml) was replaced three times per week.
From each preparation, membranes with slice cultures were randomly assigned to experimental groups. Slice cultures were used for electrophysiological recordings, fixed for immunohistochemistry or toluidine staining. For biochemical analysis, the "conditioned" culture medium was collected after 24 hr or 72 hr and stored at −80°C.
The liposome-containing medium was centrifuged.
Slice cultures were exposed for 7 to 12 days-in-vitro (DIV) to liposome-encapsulated clodronate (Liposoma B.V., Amsterdam, The Netherlands) to deplete the microglial cell population (Vinet et al., 2012). Liposomal clodronate was continuously present in the culture medium at a final concentration of 100 μg/ml from DIV 0 onward (Papageorgiou et al., 2016;Ta et al., 2019). Slice cultures were stimulated by exposures to the pro-inflammatory cytokine IFN-γ or to bacterial lipopolysaccharide (LPS). The stock solution of IFN-γ was prepared in 10 mM sterile-buffered sodium phosphate and further diluted in the culture medium to final concentrations of 100 ng/ml or 1,000 ng/ml. LPS (from Escherichia coli, serotype R515 (Re)) was ready-to-use and further diluted in the culture medium to the final concentration of 100 ng/ml. Aliquots of solutions were kept at −20°C. IFN-γ was purchased from PeproTech GmbH (Hamburg, Germany); LPS was purchased from Alexis Biochemicals (Enzo Life Sciences GmbH, Lörrach, Germany).

| Biochemical analysis of the culture medium
All enzyme-linked immunosorbent assay (ELISA) kits were purchased from R&D (R&D Systems, Inc., Minneapolis, MN, USA) and applied according to the supplier's protocol for the detection of interleukin 6 (IL-6; Cat. num. DY506) and tumor necrosis factor-alpha (TNF-α; Cat. num. DY510). Concentrations of antibodies strictly followed the suppliers' protocol. Wash buffer consisted of 0.05% Tween 20 (Merck-Millipore; Darmstadt, Germany) in PBS. Capture antibodies were diluted in PBS (pH 7.2-7.4) and the reaction plate was coated overnight. The detection antibody for TNF-α was diluted in the reagent diluent, consisting of 1% bovine serum albumin in PBS (pH 7.2-7.4); the detection antibody for IL-6 was diluted in 2% normal goat serum in reagent diluent. Ten-point standard curves were constructed from nine sequential twofold dilution steps of recombinant IL-6 (8,000 pg/ml) or TNF-α (4,000 pg/ml), and a negative control containing only reagent diluent. Samples were incubated in the coated reaction plate for 2 hr. The detection antibody was then applied for 2 hr and visualized with tetramethylbenzidine substrate solution (Moss Inc., Pasadena, USA). The development reaction was stopped with sulfuric acid, and the optical density was determined with a microplate reader (iMark Microplate Absorbance Reader, Bio-Rad Laboratories GmbH, Munich, Germany) at 450 nm (with 540 nm reference). The concentrations of TNF-α and IL-6 (both pg/ml) were estimated by using the quadratic fit.
Nitric oxide (NO) release was quantified by determining the levels of the stable metabolite nitrite using a Griess reaction-based assay that was carried out with undiluted culture medium. Ninepoint standard curves were constructed by twofold dilution steps of an 80 μΜ sodium nitrite high standard (Merck Chemicals, Darmstadt, Germany). After the addition of the Griess reagent mixture (0.05% 1-naphthylethylenediamine hydrochloride, 0.5% sulfanilamide and 2.5% orthophosphoric acid), the optical density was measured with a microplate reader at 540 nm (Bio-Rad). The molarity of NO (µM) was calculated from the standard curve using a linear fit.
Secondary antibodies were diluted in 0.2% bovine serum albumin dissolved in PBS + 0.3% Triton™ X-100. Several washing steps with PBS were conducted, for example, after blocking of unspecific binding sites or antibody applications.
For immunohistochemistry of ionized calcium-binding adapter molecule 1 (Iba1), parvalbumin-positive interneurons, glial fibrillary acidic protein (GFAP), and MHC class II (MHC-II) (see Table 1) unspecific immunoglobulin reactions were blocked for 1 hr with 10% normal serum. Primary antibodies were rabbit polyclonal anti-Iba1 (Fujifilm-WAKO Chemicals Europe GmbH, Neuss, Germany, for <5 min. Then the reaction was stopped by adding PBS (when the brown color was intense enough). Stained sections were placed on object plates and dried. Sections were then exposed to ascending ethanol series, for 10 min in xylene (Sigma-Aldrich) and finally embedded with Entellan®Neu (Merck Millipore, Schwalbach, Germany).
For toluidine blue staining (Sigma-Aldrich), sections were mounted on slides, dried and exposed to descending ethanol series, briefly rinsed in double-distilled water and then incubated in 0.1% toluidine blue working solution (pH 2.3) for 1-3 min. Thereafter, the sections were briefly rinsed in double-distilled water. Ninety-five percent ethanol with traces of glacial acetic acid was used for color differentiation of the staining. Sections were then exposed to 100% ethanol, followed by a 1:1 mixture of 100% ethanol and xylene and finally xylene for 3-10 min. Sections were embedded with Entellan®Neu (Merck Millipore, Schwalbach, Germany). Toluidine blue is a typical Nissl staining that is widely used to assess the cytoarchitecture of brain tissue.
Although toluidine blue has been also shown to stain glial cells, neurons can be clearly identified by larger cell body, nucleus, and nucleolus (Zhu, Liu, Zou, & Torbey, 2015).

| Stereological counting of microglia
The numbers of microglia (Iba1-positive cell somata) were estimated with design-based stereology recently described for hippocampal slice cultures in detail (Papageorgiou et al., 2016). In brief, we implemented the optical fractionator probe using the Stereoinvestigator® 5.65 software (MicroBrightField Europe, Delft, The Netherlands), which provides an estimator of the total particle number in a threedimensional structure. Sequential sections (total of four to seven) of each slice culture were included in the analysis. To satisfy the coefficient of sampling error (CE) < 0.1, the optimal size of the frame-associated area and grid spacing was chosen (Papageorgiou et al., 2016).
The estimated microglia number of each slice culture, N , was determined using the optical fractionator equation, that is, N = Q hsf×asf×ssf . Q was the number of the counted cells in the fractionator frameassociated area of all sections, hsf the height sampling fraction ( fractionatorheight sectionthickness ), asf the area sampling fraction ( frameassociatedarea contourareaofallsections ), ssf the section sampling fraction, that is, the interval of sections sampled through an object of interest. As we sampled every section of each slice culture, the section sampling fraction was always 1. As a result of inevitable tissue shrinkage during the staining procedures, the initial section thickness before staining (25 µm, obtained by cutting with a cryostat) had to be adjusted.
Due to very low cell numbers and irregular cell distribution in clodronate-exposed slice cultures (microglial depletion), we defined the fractionator frame-associated area as identical to the contour area of all sections in all clodronate-exposed slice cultures, that is, asf = 1. The estimated microglia number of each slice culture was determined using the optical fractionator equation. The counted cell number was corrected for tissue shrinkage (Papageorgiou et al., 2016). Initially, identification and counting of microglia were made independently by two of the authors, which resulted in similar cell numbers.

| Recordings of local field potential and [K + ] o
For electrophysiological recordings, the intact Biopore™ membrane carrying slice cultures was inserted into the recording chamber (Kann et al., 2011;Papageorgiou et al., 2016

| Data analysis and statistics
Offline analysis was performed in MatLab 11.0 or R2018B (The

| RE SULTS
Organotypic hippocampal slice cultures of the rat were maintained for 7-12 days in the incubator and exposed to liposome-encapsulated clodronate, the pro-inflammatory T lymphocyte cytokine IFN-γ (type II interferon) or bacterial LPS according to the scheme ( Figure 1a). Thereafter, the "conditioned" culture medium was used for biochemical analysis, and the slice cultures were either fixed for histology, including stereological analysis, or transferred for electrophysiological recordings to the interface recording chamber (Kann et al., 2011;Papageorgiou et al., 2016;Schneider et al., 2015). This chamber permits the continuous exchange of recording solution and ambient gas mixture as well as electrophysiological recordings.

| Microglia in situ
We characterized various features of microglia in situ, that is, in the presence of functional neuronal networks (Figures 1b-d and 2b-d).

| Hippocampal cytoarchitecture in situ
The cytoarchitecture in hippocampal slice cultures, which were stained with toluidine blue, was well-preserved. In particular, the principal cell layers, such as stratum pyramidale consisting of densely packed somata of glutamatergic pyramidal cells, were well-defined and lacked

| Cytokines and nitric oxide in situ
The levels of the pro-inflammatory cytokines IL-6 and TNF-α were very low and often beneath the detection limit when determined in the culture medium at 24 hr after the medium exchange in slice cultures (Figure 2b,c). The accumulated level of nitric oxide (NO), which was determined by the oxidation product nitrite in the culture medium at 72 hr after the medium exchange, was also low ( Figure 2d).

| Cholinergic gamma oscillations in situ
To characterize the functional integrity of local neuronal networks and astrocyte syncytia in slice cultures, we performed electrophysiological recordings.

F I G U R E 3
Recordings of cholinergic gamma oscillations in slice cultures. (a) Left, gamma oscillations were elicited by acetylcholine (2 μM) and physostigmine (400 nM) (ACh + Phy, black bar) at 34 ± 1°C. Middle, local field potential (LFP) recordings were done in stratum pyramidale of the CA3 region; DG, dentate gyrus. Right, gamma oscillations (30-70 Hz) in local neuronal networks emerge from precise mutual synaptic transmission between glutamatergic pyramidal cells (PC) and parvalbumin-positive (PV+) GABAergic interneurons that generate action potentials at 1-3 Hz and >20 Hz ("fast-spiking"), respectively. The axonal connection from PC (excitatory drive) to PV+ is not shown. Note that this is a simplified scheme. (b) Sample spectrograms of gamma oscillations in control (CTL) and clodronate-exposed ( We first applied the neurotransmitter acetylcholine and recorded local field potential responses (Figure 3a) (Papageorgiou et al., 2016;Schneider et al., 2015). Slice cultures usually show spontaneous asynchronous neuronal network activity in the absence of exogenous neurotransmitter receptor ligands (Huchzermeyer et al., 2013;Neumann et al., 1998). The continuous application of acetylcholine in slices enhances neural excitability and mimics cholinergic input to the hippocampus during exploratory behavior in vivo ). Acetylcholine reliably induced persistent gamma oscillations that had a frequency of around 40 Hz and were quite stable over time (Figure 3b,c). Notably, such cholinergic gamma oscillations in situ share many features with gamma oscillations in vivo and require both glutamatergic excitation and fast rhythmic GABAergic inhibition Kann et al., 2011).
In microglia-depleted slice cultures, gamma oscillations were still present (Figures 3b and 4a). To be able to identify even discrete im-  Schneider et al., 2019). Switches to pathological network states were also absent during gamma oscillations.
These findings suggest that the ramified and widely nonreactive microglia are not involved in the generation of gamma oscillations in situ.
The stimulus-induced [K + ] o transients increased with the stimulation intensity and showed a fast overshoot component that was followed by a prolonged undershoot component (Figure 5b,c), similar to those recorded in acute hippocampal slices and the cerebral cortex in vivo (Heinemann & Lux, 1975;Kann et al., 2005;Liotta et al., 2012). The rise of the overshoot is dominated by activity-dependent K + efflux from neurons; the fall of the overshoot and the undershoot mainly reflect K + clearance and buffering by neurons and astrocytes, which likely involves Na + /K + -ATPases, K + (K ir 4.1) channels, and gap junc- In microglia-depleted slice cultures, the stimulus-induced [K + ] o transients were unchanged (Figure 5c). There were no differences in the amplitudes of overshoot and undershoot at various stimulation intensities (Figure 5d,e). Also, switches to pathological network states such as neural burst firing or synchronized epileptic activity that would be promoted by disturbed K + homeostasis were absent during stimulation.
These findings suggest that the ramified and widely nonreactive microglia are not involved in K + clearance and buffering during robust neuronal activation in situ, neither directly nor indirectly by modulating astrocyte functions (Larsen, Stoica, & MacAulay, 2016;Varga et al., 2020).

| D ISCUSS I ON
We explored the physiological role of microglia in neuronal homeostasis and neuromodulation in postnatal cortex tissue. Our main finding is that the absence of microglia has no effects on the properties of sensitive gamma oscillations and stimulus-induced [K + ] o transients.

| Gamma oscillations and [K + ] o transients in situ and in vivo
We tested the functional integrity of the local neuronal network in the hippocampal CA3 region that is intrinsically capable of generating gamma oscillations (30-70 Hz) Kann et al., 2011). Gamma oscillations require precise chemical and electrical synaptic transmission between excitatory pyramidal cells and inhibitory interneurons (Colgin, 2016;).
Among the latter, parvalbumin-positive, GABAergic basket cells, which generate action potentials at >20 Hz ("fast-spiking") and feature unique biophysical and bioenergetic properties, have a key role in the rhythm generation (Gulyás et al., 2010;Kann et al., 2014).
They share key features with gamma oscillations in hippocampal acute slices and the hippocampus in vivo, such as oscillation generation in the CA3 region, average frequency of around 40 Hz and reversal of the local field potential between the pyramidal cell layer (stratum pyramidale) and the apical dendritic layer (stratum radiatum) Kann et al., 2011;Vodovozov et al., 2018).
Gamma oscillations reflect the overall increase in excitatory and inhibitory membrane currents in neurons .
Gamma oscillations are also modulated by a variety of homeostatic factors, including the intracellular pH (Stenkamp et al., 2001) and the extracellular levels of potassium ions, ATP, metabolites such as lactate and pyruvate, and amino acids such as glycine (Galow et al., 2014;LeBeau et al., 2002;Schulz et al., 2012;Vodovozov et al., 2018).
Therefore, gamma oscillations provide a sensitive readout of even discrete neuronal network dysfunction associated with disturbances in neuronal ion and energy homeostasis and/or neuromodulation.
We report that the absence of microglia did not affect the properties of sensitive neuronal gamma oscillations and stimulus-in- We applied local field potential recordings that integrate mainly synaptic transmembrane currents and have a spatial reach of a few hundred micrometers around the electrode tip (Einevoll, Kayser, Logothetis, & Panzeri, 2013). Therefore, our data on gamma oscillations on the network level cannot exclude that the absence of microglia might have contributed to subtle alterations in intrinsic membrane properties, synaptic connectivity, and/or neurotransmission of glutamatergic and GABAergic neurons (Antonucci et al., 2012;Ji, Akgul, Wollmuth, & Tsirka, 2013;Parkhurst et al., 2013;Pribiag & Stellwagen, 2013). However, the presence of regular gamma oscillations and stimulus-induced [K + ] o transients argues for widely intact neuronal and astrocytic functions that are crucial for cortical information processing (Colgin, 2016;Kann et al., 2014;Rasmussen, O'Donnell, Ding, & Nedergaard, 2020).
Our data support the biological concept that nonreactive (surveying) microglia are dispensable for neuronal homeostasis, signaling, and rhythm generation in the healthy postnatal brain. This evolutionary feature might permit microglia to use the enormous phenotypic flexibility, which often includes local proliferation and migration, in response to any homeostatic imbalance and pathology within the CNS.

D ECL A R ATI O N O F TR A N S PA R EN C Y
The authors, reviewers and editors affirm that in accordance to the policies set by the Journal of Neuroscience Research, this manuscript presents an accurate and transparent account of the study being reported and that all critical details describing the methods and results are present.

ACK N OWLED G M ENTS
The authors thank Michal Schwartz, Amit Agarwal, Thomas Blank, and Marco Prinz for helpful discussions.

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

AUTH O R CO NTR I B UTI O N S
All the authors take responsibility for the integrity of the data and the accuracy of the data analysis. Project Administration, A.L. and O.K.

PE E R R E V I E W
The peer review history for this article is available at https://publo ns.com/publo n/10.1002/jnr.24689.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are available from the corresponding author upon reasonable request.