Scavenging Tumor Necrosis Factor α Does Not Affect Inhibition of Dentate Granule Cells Following In Vitro Entorhinal Cortex Lesion

Neurons that lose part of their afferent input remodel their synaptic connections. While cellular and molecular mechanisms of denervation-induced changes in excitatory neurotransmission have been identified, little is known about the signaling pathways that control inhibition in denervated networks. In this study, we used mouse entorhino-hippocampal tissue cultures of both sexes to study the role of the pro-inflammatory cytokine tumor necrosis factor α (TNFα) in denervation-induced plasticity of inhibitory neurotransmission. In line with our previous findings in vitro, an entorhinal cortex lesion triggered a compensatory increase in the excitatory synaptic strength of partially denervated dentate granule cells. Inhibitory synaptic strength was not changed 3 days after the lesion. These functional changes were accompanied by a recruitment of microglia in the denervated hippocampus, and experiments in tissue cultures prepared from TNF-reporter mice [C57BL/6-Tg(TNFa-eGFP)] showed increased TNFα expression in the denervated zone. However, inhibitory neurotransmission was not affected by scavenging TNFα with a soluble TNF receptor. In turn, a decrease in inhibition, i.e., decreased frequencies of miniature inhibitory postsynaptic currents, was observed in denervated dentate granule cells of microglia-depleted tissue cultures. We conclude from these results that activated microglia maintain the inhibition of denervated dentate granule cells and that TNFα is not required for the maintenance of inhibition after denervation.


Introduction
The balance of excitation and inhibition is essential for the proper function and for information processing in cortical circuits [1][2][3]. While feedforward and feedback circuits dynamically match recruited inhibition to afferent excitation and local network activity, synaptic plasticity maintains this balance over longer periods of time [4]. Homeostatic synaptic plasticity-a synaptic mechanism that is based on negative feedback mechanismsis considered to play an important role in this context [5,6]. Work from the past two decades has identified several cellular and molecular mechanisms that mediate and modulate the ability of neurons to express the homeostatic plasticity of excitatory and inhibitory neurotransmission [7][8][9]. Nevertheless, the biological significance of homeostatic synaptic changes that occur under pathological conditions such as brain injury and neurodegeneration remains elusive.
Neurons that are not directly affected by a lesion but lose part of their afferent excitation following brain injury increase their excitatory synaptic strength in a compensatory manner [10][11][12][13][14][15]. Interestingly, the downscaling of inhibitory neurotransmission is not

Ethics Statement
Mice were maintained in a 12 h light/dark cycle with food and water available ad libitum. Every effort was made to minimize the distress and pain of animals. Experimental procedures (X-17/07K; X-17/09C) were performed according to German animal welfare legislation and approved by the appropriate animal welfare committee and the animal welfare officer of the University of Freiburg.

Preparation of Tissue Cultures
Organotypic entorhino-hippocampal tissue cultures were prepared at postnatal day 4-5 from mice of either sex as previously described [29]. The tissue cultures were transferred onto porous (0.4 µm pore size, hydrophilic PTFE) cell culture inserts with 30 mm diameter (Merck/Millipore, Darmstadt, Germany, Cat# PICM0RG50) for cultivation. The culturing medium consisted of 50% (v/v) minimum essential medium (MEM), 25% (v/v) basal medium eagle (BME), 25% (v/v) heat-inactivated normal horse serum (NHS), 2 mM GlutaMAX, 0.65% (w/v) glucose, 25 mM HEPES buffer solution, 0.1 mg/mL streptomycin, 100 U/mL penicillin and 0.15% (w/v) bicarbonate. The pH of the culturing medium was adjusted to 7.30 and tissue cultures were incubated for at least 18 days at 35 • C in a humidified atmosphere with 5% CO 2 . The culturing medium was replaced thrice a week at 2and 3-day intervals. After the lesion, the medium was not replaced for 3 days until further experimental assessment.

Entorhinal Cortex Lesion
In order to assure a complete and reproducible denervation of granule cells in the dentate gyrus, the entorhinal cortex was transected and removed from the culturing insert

Entorhinal Cortex Lesion
In order to assure a complete and reproducible denervation of granule cells in the dentate gyrus, the entorhinal cortex was transected and removed from the culturing insert with a sterile scalpel blade ( Figure 1A,B). All experiments were performed 3 days after the lesion. and group data of AMPA receptor-mediated miniature excitatory postsynaptic currents (mEPSCs) recorded from dentate granule cells of non-denervated controls and age-and time-matched denervated tissue cultures 3 days after the lesion (control, n = 17 cells from 4 cultures; ECL 3d, n = 18 cells from 4 cultures; Mann-Whitney test; ** p < 0.01 NS, non-significant difference). For these graphs and for the rest of the graphs, values represent mean ± standard error of the mean (s.e.m.), and gray dots indicate individual data points.

Pharmacology
Organotypic entorhino-hippocampal tissue cultures (≥18 days in vitro) were treated with mouse recombinant soluble tumor necrosis factor receptor 1 (sTNFR1 10 μg/mL; R&D systems, Minneapolis, MN, USA, Cat# 425-R1-050), 20 min before the lesion and for 3 days. For the depletion of microglia in tissue cultures, PLX3397 (50 nM in DMSO; Axon MedChem, Groningen, Netherlands, Cat# 2501) was added to the culturing medium immediately after preparation and during the whole cultivation period. The respective control tissue cultures were treated with equal volume of dimethyl sulfoxide (DMSO).

Pharmacology
Organotypic entorhino-hippocampal tissue cultures (≥18 days in vitro) were treated with mouse recombinant soluble tumor necrosis factor receptor 1 (sTNFR1 10 µg/mL; R&D systems, Minneapolis, MN, USA, Cat# 425-R1-050), 20 min before the lesion and for 3 days. For the depletion of microglia in tissue cultures, PLX3397 (50 nM in DMSO; Axon MedChem, Groningen, Netherlands, Cat# 2501) was added to the culturing medium immediately after preparation and during the whole cultivation period. The respective control tissue cultures were treated with equal volume of dimethyl sulfoxide (DMSO).

Post-Hoc Identification of Recorded Neurons
For mEPSC recordings, biocytin-containing internal solution was used as described above. After recording, tissue cultures were fixed in a solution of 4% (w/v) paraformaldehyde (PFA) and 4% (w/v) sucrose in 0.01 M PBS for 1 h at room temperature. After washing with 0.01 M PBS, the fixed tissue was incubated for 1 h at room temperature in a blocking solution consisting of 10% (v/v) normal goat serum (NGS) and 0.5% (v/v) Triton X-100 in 0.01 M PBS. Biocytin-filled cells were stained with Alexa-633 conjugated Streptavidin (ThermoFisher Scientific, Waltham, MA, USA, Cat# S21375) in a dilution of 1:1000 in 0.01 M PBS with 10% NGS and 0.1% Triton X-100 overnight at 4 • C. DAPI staining was used to visualize cytoarchitecture (1:2000; in 0.01 M PBS for 15 min). Slices were then washed 3 times in 0.01 M PBS, transferred and mounted onto glass slides with anti-fading mounting medium (DAKO) for visualization.

Quantification and Statistics
Single cell recordings were analyzed using Clampfit 11 of the pClamp11 software package (Molecular Devices, San Jose, CA, USA). mEPSC and mIPSC properties were analyzed using the automated template search tool for event detection, which detects the events in an unbiased fashion. Confocal image stacks of heterozygous C57BL/6-Tg(TNFa-eGFP) cultures were processed and analyzed using the Fiji image processing package (available at http://imagej.net; accessed on 17 November 2021). From each stack, maximum intensity projection of 10 images was performed. The molecular layer of the dentate gyrus was manually defined as the region of interest (ROI) and the mean fluorescence intensity of each ROI was measured. For the assessment of microglia numbers, single plane images were analyzed. The molecular layer of the dentate gyrus was manually defined as the region of interest (ROI). After background subtraction (30 px), cell numbers were measured using the 'Analyze Particles' function of Fiji. Analysis was performed by the investigator blind to experimental conditions. Statistical comparisons were carried out using GraphPad Prism 7 (GraphPad software, San Diego, CA, USA). Non-parametric tests were performed (Mann-Whitney test and Kruskal-Wallis test followed by Dunn's post hoc test) for the analysis of all experiments in the present study.

Digital Illustrations
Figures were prepared using the Affinity Designer (Serif Europe, Nottingham, UK) and the Adobe Photoshop (Adobe, San Jose, CA, USA) graphics software. Image brightness and contrast were adjusted. Figure 1A was created with BioRender (www.biorender.com; accessed on 17 November 2021).

Partial Denervation of Dentate Granule Cells Induces a Compensatory Adjustment of Excitatory Synaptic Strength
The denervation-induced homeostatic synaptic plasticity of dentate granule cells was probed in mouse organotypic entorhino-hippocampal tissue cultures (≥18 days in vitro) as described before [11,12,19,23,24]. Slice cultures were transected from the rhinal fissure to the hippocampal fissure using a sterile scalpel blade ( Figure 1A,B). To ensure complete and reproducible separation of the entorhinal cortex from the hippocampus, the entorhinal cortex was removed in every denervation experiment from the culturing insert. This procedure does not affect the target neurons in the dentate gyrus directly but leads to a reproducible partial denervation of distal dentate granule cell dendrites in the molecular layer of the dentate gyrus ( Figure 1C). Three days after the lesion, individual dentate granule cells in the suprapyramidal blade of the dentate gyrus were patched and AMPA receptor-mediated miniature excitatory postsynaptic currents (mEPSCs) were recorded ( Figure 1D). The amplitude of these events reflects excitatory synaptic strength. Consistent with our previous findings [11,12,19,23,24], a homeostatic increase in excitatory synaptic strength, i.e., increased mEPSC amplitudes, was observed 3 days after denervation ( Figure 1E). These results confirm once more that in vitro denervation of dentate granule cells induces a robust compensatory strengthening of excitatory synapses.

TNFα Does Not Affect Inhibitory Neurotransmission onto Denervated Dentate Granule Cells
Our previous work showed that denervation-induced excitatory synaptic strengthening depends on TNFα [23,24]. However, the role of TNFα for inhibitory neurotransmission onto partially denervated dentate granule cells has not yet been addressed.
Before assessing inhibitory neurotransmission, we confirmed that entorhinal cortex lesion triggers Tnfα gene expression in the denervated zone [23]. In these experiments, we used tissue cultures prepared from a TNF-reporter mouse line, which expresses enhanced Cells 2021, 10, 3232 6 of 12 green fluorescent protein (GFP) under the control of the TNF promoter [28]. Indeed, a significant increase in the GFP signal was observed in the molecular layer of denervated tissue cultures, thus confirming increased Tnfα gene expression in the denervated zone following in vitro entorhinal cortex lesion (Figure 2A,B). Next, GABA receptor-mediated miniature inhibitory postsynaptic currents (mIPSCs) were recorded from individual control and denervated dentate granule cells of another set of wild-type cultures ( Figure 3A). Three days after denervation, no significant differences in inhibitory synaptic strength were observed between the two groups ( Figure 3B), thus corroborating our earlier findings on unaltered inhibitory neurotransmission following in vitro denervation [19].
We then used soluble TNFα receptor (sTNFR; 10 µg/mL) to scavenge TNFα. Our previous work showed that denervated dentate granule cells do not show increased excitatory synaptic strength 3 days after denervation in the presence of sTNFR [23]. sTNFR was added shortly before the entorhinal cortex lesion to the culturing medium for 3 days. Figure 3B shows that mIPSC amplitudes and frequencies were comparable in control and lesioned sTNFR-treated tissue cultures 3 days after the lesion ( Figure 3B). We conclude Next, GABA receptor-mediated miniature inhibitory postsynaptic currents (mIPSCs) were recorded from individual control and denervated dentate granule cells of another set of wild-type cultures ( Figure 3A). Three days after denervation, no significant differences in inhibitory synaptic strength were observed between the two groups ( Figure 3B), thus corroborating our earlier findings on unaltered inhibitory neurotransmission following in vitro denervation [19].    We then used soluble TNFα receptor (sTNFR; 10 µg/mL) to scavenge TNFα. Our previous work showed that denervated dentate granule cells do not show increased excitatory synaptic strength 3 days after denervation in the presence of sTNFR [23]. sTNFR was added shortly before the entorhinal cortex lesion to the culturing medium for 3 days. Figure 3B shows that mIPSC amplitudes and frequencies were comparable in control and lesioned sTNFR-treated tissue cultures 3 days after the lesion. We conclude from these results that scavenging TNFα both in unlesioned control cultures and following denervation does not affect mIPSC properties of dentate granule cells.

Microglia Maintain Inhibition of Partially Denervated Dentate Granule Cells
To confirm and extend these findings, we tested for the role of microglia on inhibitory neurotransmission in our experimental setting, also considering that microglia are a major source of TNFα in the central nervous system [25]. Immunostainings for the microglial marker Iba1 revealed a higher number of microglial cells in the denervated zone, suggesting an involvement of microglia in denervation-induced plasticity ( Figure 4A

Microglia Maintain Inhibition of Partially Denervated Dentate Granule Cells
To confirm and extend these findings, we tested for the role of microglia on inhibitory neurotransmission in our experimental setting, also considering that microglia are a major source of TNFα in the central nervous system [25]. Immunostainings for the microglial marker Iba1 revealed a higher number of microglial cells in the denervated zone, suggesting an involvement of microglia in denervation-induced plasticity ( Figure 4A,B; [30]).  To test for the role of microglia on inhibitory neurotransmission, microglia were depleted from tissue cultures using the colony stimulating factor 1 receptor (CSF1R) antagonist PLX3397 [31][32][33]. Tissue cultures were exposed to 50 nM PLX3397 right after preparation and kept in the same concentration of PLX3397 until and during experimental assessment. Using this approach, the vast majority of microglia were successfully depleted from tissue cultures ( Figure 5A). We previously showed that the depletion of microglia from tissue cultures does not affect cell viability and basic functional and structural properties of neurons [33]. Consistent with this suggestion, no changes in the mean series resistance and capacitance of dentate granule cells were observed between the groups (R control = 7.304 MΩ ± 0.45: R PLX3397 = 7.959 MΩ ± 0.46; C control = 20.000 pF ± 1.89: C PLX3397 = 17.530 pF ± 1.49). Once more, mIPSCs were recorded from individual dentate granule cells and we found no significant differences in the amplitudes and frequencies between non-depleted control and denervated tissue cultures ( Figure 5B,C). While the depletion of microglia had no apparent effects on baseline mIPSC properties, a significant reduction in mIPSC Cells 2021, 10, 3232 8 of 12 frequencies was observed 3 days after denervation in the microglia-depleted tissue cultures ( Figure 5B,C). We conclude from these results that microglia maintain inhibitory synaptic strength of dentate granule cells following in vitro partial denervation. from tissue cultures does not affect cell viability and basic functional and structural properties of neurons [33]. Consistent with this suggestion, no changes in the mean series resistance and capacitance of dentate granule cells were observed between the groups (Rcontrol = 7.304 MΩ ± 0.45: RPLX3397 = 7.959 MΩ ± 0.46; Ccontrol = 20.000 pF ± 1.89: CPLX3397 = 17.530 pF ± 1.49). Once more, mIPSCs were recorded from individual dentate granule cells and we found no significant differences in the amplitudes and frequencies between non-depleted control and denervated tissue cultures ( Figure 5B,C). While the depletion of microglia had no apparent effects on baseline mIPSC properties, a significant reduction in mIPSC frequencies was observed 3 days after denervation in the microglia-depleted tissue cultures ( Figure 5B,C). We conclude from these results that microglia maintain inhibitory synaptic strength of dentate granule cells following in vitro partial denervation.

Discussion
The results of this study show that scavenging TNFα for 3 days with sTNFR does not affect inhibitory synapses, i.e., mIPSC properties, in cultured dentate granule cells, both under control conditions and following partial denervation. However, a denervationinduced reduction in inhibition, i.e., decreased mean mIPSC frequency, is observed when microglia are depleted from tissue cultures. This finding is consistent with a compensatory adjustment of inhibition following partial denervation, i.e., decreased inhibition in response to the loss of afferent excitation. Since this homeostatic adjustment of inhibition did not occur in the presence of microglia or when TNFα was scavenged, we conclude that microglia maintain the inhibition of partially denervated dentate granule cells in a TNFαindependent manner. The ability of neurons to compensate for perturbations in afferent excitation is considered fundamental for normal brain function [5,6]. During the past few decades, several molecular mechanisms, which mediate homeostatic synaptic plasticity, have been identified [7][8][9]. It is now well-established that activity deprivation triggers the synaptic upscaling of excitatory neurotransmission and downscaling of synaptic inhibition [34][35][36][37]. However, the biological significance of compensatory changes at excitatory and inhibitory synapses remains unclear. For example, it is currently unknown whether homeostatic synaptic mechanisms support or obstruct the network integration of newly born dentate granule cells [38] and the recovery of synaptic connectivity after brain injury [39]. Likewise, the mechanisms that coordinate homeostatic changes at excitatory and inhibitory synapses at the single cell level are subject to further investigation.
Evidence has been provided that distinct mechanisms regulate the homeostasis of excitatory and inhibitory synapses. For example, CaMKIV-/pCREB-mediated signaling pathways mediate compensatory changes of excitatory synaptic strength (and intrinsic cellular properties) but not inhibitory neurotransmission [40]. Similarly, Clptm1 affects GABA A receptors without changing AMPAR-mediated currents [41]. Conversely, the proinflammatory cytokine TNFα has been linked to the ability of neurons to express both excitatory and inhibitory synaptic plasticity [21,22,42], thus indicating that TNFα could be a key mediator of coordinated excitatory and inhibitory homeostatic synaptic changes.
Indeed, our previous work showed that TNFα maintains increased excitatory synaptic strength following an in vitro entorhinal cortex lesion [23,24]. Specifically, the initial denervation-induced compensatory increase in excitatory synaptic strength returned to baseline 3 days post lesion when TNFα was scavenged with 10 µg/mL sTNFR [23]. These results suggest that TNFα is required for the maintenance of denervation-induced synaptic strengthening. In a follow-up study, we assessed denervation-induced homeostatic changes of inhibitory neurotransmission and found no significant changes after the lesion [19]. Accordingly, we theorized that TNFα could be one of the major homeostatic factors that maintain increased excitatory and unchanged inhibitory synaptic neurotransmission in denervated neural networks. While we were able to reproduce our earlier findings on denervation-induced synaptic changes, the results of the present study showed that scavenging TNFα with soluble TNFR (sTNFR) does not trigger the downscaling of inhibitory synapses after denervation. Although we cannot exclude that TNFα-dependent changes in inhibitory neurotransmission may occur at a different time point after denervation, the results of the present study demonstrate that TNFα per se is not a key mediator of synchronous homeostatic changes at excitatory and inhibitory synapses, at least not in the denervated dentate gyrus 3 days after in vitro entorhinal cortex lesion.
Possible concentration-dependent effects of TNFα must be considered in this context [43][44][45]. For instance, low concentrations of TNFα reduce inhibitory neurotransmission in dissociated cultures [21,22], while high concentrations of TNFα-as seen after in vivo administration of bacterial lipopolysaccharide (LPS)-seem to enhance inhibitory neurotransmission [46]. While it is not trivial to measure the exact concentration of endogenous TNFα at synaptic sites, denervation seems to be a mild TNFα-inducing stimulus, specifically when compared to the exposure of tissue cultures to LPS [28]. It is therefore interesting to speculate that very low concentrations of TNFα may affect excitatory but not inhibitory neurotransmission. At a higher concentration, TNFα reduces inhibition [21,22], while very high concentrations of TNFα might increase synaptic inhibition [46]. This suggestion is supported by our previous work on the dose-dependent effects of TNFα in excitatory synaptic plasticity [43], where low concentrations of TNFα promoted LTP-possibly by reducing inhibition-and high concentrations blocked the ability of neurons to express LTP-possibly by increasing inhibition. Apparently, more work is required to clarify the concentration-dependent effects of TNFα on inhibitory neurotransmission. It will also be important to test for input-/synapse-specific effects of TNFα on dendritic vs. somatic inhibition in this context, since in the present study increased TNFα expression was observed mainly in the denervated molecular layer, where the distal dendrites of denervated granule cells are located [11,23,39].
Considering that distinct mechanisms seem to regulate excitatory and inhibitory neurotransmission, we tested for the role of microglia in our experimental setting. In line with previous in vivo work [30], increased numbers of microglia were observed in the molecular layer of denervated tissue cultures. Interestingly, when the majority of microglia was depleted from tissue cultures using an established protocol, i.e., PLX3397 [31][32][33], a homeostatic decrease in inhibition, i.e., a reduction in mIPSC frequencies, was demonstrated in partially denervated dentate granule cells. Notably, no differences in mIPSC properties were observed between non-depleted and microglia-depleted tissue cultures under baseline conditions, indicating that the presence of microglia is not required for dentate granule cells to obtain and maintain their inhibitory synaptic set-point under control conditions. However, the denervation-induced activation of microglia seems mandatory for the maintenance of inhibition following denervation. These findings suggest that microglia prevent the synaptic downscaling of inhibition in partially denervated neural networks. It is unlikely that TNFα mediates this function of microglia, since sTNFR (10 µg/mL), which scavenges TNFα (regardless of its source), did not affect inhibitory neurotransmission under control conditions and 3 days following in vitro denervation.
Nevertheless, it should be noted in this context that astrocytic and neuronal sources of TNFα have been reported [47][48][49]. Our experiments using TNF-reporter mice did not provide any evidence for a neuronal source of Tnfα, since no changes in GFP signal were observed in the granule cell layer. However, at this point we cannot fully exclude astrocytic sources of TNFα that may have escaped our detection in the denervated molecular layer, also considering our previous work that indicated changes in astrocytic TNFα following an entorhinal cortex lesion in vitro [23]. We are confident that future work in organotypic tissue cultures will support the identification of the cellular and molecular mechanisms through which microglia and astrocytes coordinate homeostatic changes in synaptic excitation/inhibition balance under physiological conditions and following partial denervation. Institutional Review Board Statement: Experimental procedures (X-17/07K; X-17/09C) were approved by the appropriate animal welfare committee and the animal welfare officer of Freiburg University.

Informed Consent Statement: Not applicable.
Data Availability Statement: All data obtained during the present study are available from the corresponding author upon reasonable request.