Bacterial peptidoglycan signalling in microglia: Activation by MDP via the NF-κB/MAPK pathway

Bacterial peptidoglycan (PGN) fragments are commonly studied in the context of bacterial infections. However, PGN fragments recently gained recognition as signalling molecules from the commensal gut microbiota in the healthy host. Here we focus on the minimal bioactive PGN motif muramyl dipeptide (MDP), found in both Gram-positive and Gram-negative commensal bacteria, which signals through the Nod2 receptor. MDP from the gut microbiota translocates to the brain and is associated with changes in neurodevelopment and behaviour, yet there is limited knowledge about the underlying mechanisms. In this study we demonstrate that physiologically relevant doses of MDP induce rapid changes in microglial gene expression and lead to cytokine and chemokine secretion. In immortalised microglial (IMG) cells, C-C Motif Chemokine Ligand 5 (CCL5/RANTES) expression is acutely sensitive to the lowest physiologically prevalent dose (0.1 µg/ml) of MDP. As CCL5 plays an important role in memory formation and synaptic plasticity, microglial CCL5 might be the missing link in elucidating MDP-induced alterations in synaptic gene expression. We observed that a higher physiological dose of MDP elevates the expression of cytokines TNF-α and IL-1β, indicating a transition toward a pro-inflammatory phenotype in IMG cells, which was validated in primary microglial cultures. Furthermore, MDP induces the translocation of NF-κB subunit p65 into the nucleus, which is blocked by MAPK p38 inhibitor SB202190, suggesting that an interplay of both the NF-κB and MAPK pathways is responsible for the MDP-specific microglial phenotype. These findings underscore the significance of different MDP levels in shaping microglial function in the CNS and indicate MDP as a potential mediator for early inflammatory processes in the brain. It also positions microglia as an important target in the gut microbiota-brain-axis pathway through PGN signalling.


Introduction
It is now widely recognized that the gut microbiota is critically involved in shaping brain development, function, and behaviour [1].Studies indicate that microglia play a significant role in orchestrating the interplay between gut microbes and the brain [2,3].In addition, microglia are important for neurodevelopment, homeostasis and synaptic refinement later in life [4,5].
Although cytokines and chemokines produced by microglia are responsible for the primary neuroinflammatory response in the brain, these proteins also positively regulate synapse formation and plasticity [6].
In germ-free (GF) mice, which are born and raised under strict sterile conditions, microglia remain in their immature highly ramified state and exhibit a premature transcriptomic profile [1].These microglia exhibit an impaired inflammatory response to lipopolysaccharide (LPS) challenge, resulting in decreased production of pro-inflammatory cytokines such as interleukin (IL)-1β, IL-6 and tumor necrosis factor-α (TNF-α) [2].IL-1β and TNF-α are major regulators of excitatory synapses by mediating glutamatergic signalling, thereby regulating synaptic plasticity.An opposing effect is evident with IL-6 signalling, weakening basal excitatory neurotransmission [7].C-C Motif Chemokine Ligand 5 (CCL5) has been identified as an important regulator of glutamate and glucose aerobic metabolism that affects synapse formation and memory [8].In microglia, NF-κB is one of the master regulators of gene expression, including cytokines and chemokines.Upon recognition of bacterial molecules, signal transduction pathways such as the mitogen-activated protein kinase (MAPK) are activated, leading to the translocation of NF-κB proteins into the nucleus and cytokine expression [9].Microglia reactivity to pathogens, injury, disease and aging is well described [10,11]; however, elucidation of how commensal gut microbiota signals influence microglia activity remains incomplete.
Recent studies have implicated the central activation of germline-encoded pattern-recognition receptors (PRRs) in the innate immune system as potential key regulators of gut microbiotabrain interactions [12][13][14][15].These receptors recognize conserved microbial molecular signatures such as bacterial surface molecules (e.g.peptidoglycans, PGNs).Traditionally, the investigation of bacterial cell wall components such as PGN entering the brain, has centred on compromised blood-brain barrier (BBB) function linked to infections [16].Recent insights into the complexity of the human microbiota and its impact on health have prompted a reassessment of host-microbe interactions.The gut microbiota, housing trillions of indigenous bacteria, generates a diverse 'peptidoglycome' that can systemically reach peripheral organs, including the brain, indicating broader implications for physiological host processes [17,18].Peptidoglycan (PGN) is a fundamental component in the cell walls of almost all bacteria.The PGN backbone comprises alternating sugar residues-β-(1-4)-linked N-acetylglucosamine (GlcNAc) and N-acetylmuramic acid (MurNAc)-cross-linked by short peptides containing twoto-five amino acids.Remarkably, PGN undergoes continuous remodelling throughout the bacterial life cycle, leading to the shedding of fragments into the environment, a process known as PGN turnover.Major families of pattern recognition receptors (PRRs) that specifically identify PGN fragments include PGN-recognition proteins (PGRPs; Pglyrp1-4) and intracellular NOD-like receptors (NLRs; Nod1 and Nod2).Extensive research has been conducted to understand the structural requirements for Nod1 and Nod2 recognition of PGN.For instance, Nod2 activation necessitates an intact MurNAc sugar ring with an attached dipeptide moiety (L-Ala-D-Glu or L-Ala-D-isoGln).Consequently, Nod2 is strongly activated by muramyl dipeptide (MDP), the minimally bioactive PGN fragment present in both Gram-positive and Gram-negative bacteria [19].Recent studies have demonstrated that Nod2 is expressed in various brain regions, including the prefrontal cortex, striatum and hypothalamus, with particularly high abundance in microglia [20].It has become evident in recent years that MDP from commensal gut microbiota can translocate to the systemic circulation and reach the brain.Optimal levels of MDP are crucial for healthy host neurodevelopment [19,21].
Therefore, understanding the regulatory role of peptidoglycan (PGN) fragments, such as MDP, in the brain is essential.Importantly, not many studies have focused on MDP as a signalling molecule released by commensal gut microbiota during physiological concentrations.
In the present study, we investigate the impact on microglial function of physiological (0.1 and 1 µg/ml) and high (10 µg/ml) doses of synthetic MDP, mimicking its analogue present in commensal gut microbiota.We demonstrate that low levels of MDP from non-pathogenic bacteria induce CCL5 expression, whereas upper physiological and high MDP levels shift microglia toward a pro-inflammatory phenotype by engaging the NF-κB and MAPK pathways.This revelation might position microglia as the anchor point in the gut microbiota-brain-axis pathway involving PGN signalling.We focus on several cytokines and chemokines reported to be important for regulating neuronal connectivity and their underlying pathway of expression in both immortalized microglial cells (IMG) and primary microglia.

Animals
C57BL/6NTac mice (Taconic) were bred at the Comparative Medicine Department at Karolinska University Hospital, Sweden.Animals were maintained in a pathogen-free and climatecontrolled environment with regulated 12-h light/dark cycles and had access to food and water ad libitum.
The animal experiments were performed according to the rules of the Swedish National Board of Laboratory Animals, the European Community Council Directive (86/609/ECC) and the local ethics committee of Stockholm under the ethical permit 9328-2019.

Primary microglia culture preparation
Primary microglia cultures were prepared from adult mouse brain as previously described [22].
2-month-old C57BL/6NTac mice were anaesthetised and perfused with cold PBS followed by collection of the brains, which were minced with a pipette and dissociated enzymatically in 5 ml L15 medium containing papain (Worthington, LS003126; 1:100) and DNase I (Roche, 10104159001, 0.2 mg/ml).The resulting brain homogenate solution was incubated in 37°C for 20 min.After 10 min, the brain homogenates were triturated using a pipette for 10 min to create a single cell suspension.The enzymatic activity was blocked by adding 20 ml cold HBSS and the resulting homogenate was passed through a 100 µm cell strainer and centrifuged at 300 g, 5 min, 4°C.The pellets were resuspended in 20 ml 37% isotonic Percoll (Sigma-Aldrich, P1644) in HBSS and centrifuged at 800 g (acceleration 4, deceleration 0) for 10 min at 4°C.The supernatant containing the myelin layer on the top was discarded.The pellets containing mixed glial cells were resuspended in DMEM/F12 complete medium (microglia medium) supplemented with 10% FBS, 20 ng/ml recombinant mouse M-CSF (R&D Systems, 416-ML), 2 mM l-glutamine (Sigma-Aldrich, G7513), 100 U/ml penicillin, and 100 μg/ml streptomycin (Sigma-Aldrich, P4458).Mixed glial cells were cultured for 14 days with medium change 3 times per week.After reaching confluency the cells were detached with 0.05% Trypsin-EDTA in 37°C for 10 min.Finally, microglia were isolated from the mixed glial culture using anti-CD11b magnetic beads (MiltenyiBiotec, 130-049-601) according to the manufacturer's protocol.
IMG cells recapitulate many features of primary microglial culture, as they are CD11b + and F4/80 + and express other microglia-specific markers such as Iba1, TMEM119, and P2Ry12 [23][24][25].Furthermore, IMG cells respond to a lipopolysaccharide (LPS) challenge similarly to primary microglia, producing comparable levels of levels of TNF-α and other cytokines/chemokines [25].Taken together, these previous studies using IMG cells suggest the cell line to be a suitable microglial cell culture model for our study.
Cells were sub-cultured as recommended by the supplier and not for more than 10 passages after thawing to reduce genetic drift and increase reproducibility.For RT-qPCR, cells were seeded at a density of 1x10 5 in 6-well tissue culture-treated plates in 2 ml of media.Cells used for ELISA were cultured in 6-well or 24-well plates (1x10 5 cells/well or 3x10 4 cells/well) in 2ml or 500µl media, respectively.Immunocytochemistry was performed at a seeding concentration of 5x10 4 cells/well in black 24-well plates with flat and clear bottoms for high throughput microscopy (ibidi; cat# 82426).For treatment with MDP or MAPK inhibitor, cells were kept in cell culture media.MDP was diluted in DMEM high glucose media (Sigma; cat# D6546).DMEM high glucose media was added in the control wells.

MDP treatment
75% of healthy human donors have serum levels of MDP ranging from 0.16-1 µg/ml, whereas specific pathogen-free (SPF) mice with an intact microbiota have MDP levels of around 0.2 µg/ml [26,27], and PGN levels of 0.6 ng/mg were detected in the cerebellum [12].Slight changes in MDP concentrations might occur due to dysbiosis in the gut microbiota community.

MAPK p30 inhibitor treatment
The MAPK p38 inhibitor SB202190 (SB20) (Sigma; cat#S7076-5MG) was used to assess the downstream signalling pathway of MDP.SB20 was resuspended in DMSO and diluted in DMEM high glucose media (Sigma; cat# D6546).Control and MDP treated cells received equivalent DMSO levels compared to SB20 dilutions.After 1 h pre-incubation of IMG cells with 10 µM SB20, cells were exposed to 10 µg/ml MDP for 3 h.RNA expression was measured using RT-qPCR.To further investigate a potential interplay of both NF-κB pathway and MAPK p38 pathway, cells were incubated with 10 µM SB20 for 1 h, followed by 1 h incubation with 10 µg/ml MDP.Nuclear translocation of NF-κB p65 protein was assessed by confocal imaging.

Quantitative real-time polymerase chain reaction (RT-qPCR)
RT-qPCR was used to evaluate the gene expression patterns of IMG and primary microglia cells in response to MDP.For the MDP time curve, IMG cells were incubated with 10 µg/ml MDP for 90 min, 3 h or 24 h, respectively.The concentration-dependent effects of MDP were assessed using 0.1 µg/ml, 1 µg/ml, and 10 µg/ml doses for 3 h and 24 h, respectively.All conditions were measured in 6 biological and 3 technical replicates.The total RNA was isolated using the RNeasy® Mini Kit as per the manufacturer's instructions (Qiagen; cat# 74104).RNA concentration and purity were determined using the ND-1000 spectrophotometer (Nanodrop; ThermoFisher).250 ng of RNA were reverse transcribed into cDNA using the iScript® cDNA synthesis kit (Bio-Rad; cat#1708890).Gene expression was analysed using the CFX384 Touch Real-Time PCR Detection System (Bio-Rad).Primers were designed by using Primer-BLAST web-based software (http://www.ncbi.nlm.nih.gov/tools/primer-blast/). Primer sequences and gene accession numbers are listed in Supplementary Table 1. 5 µl SYBR® Green Supermix (SYBR® Green I dye; Bio-Rad; cat# 1725124) together with 0.5 µM of each primer, 3 µl DNA were adjusted to a 10 µl reaction with nuclease-free water.Each condition was measured in 6 biological and 3 technical replicates.Thermocycling steps are shown in Table 1.Relative gene expression was analyzed using the 2− ΔΔCt method.

Selection of Stable Reference Genes
Normalization with reference genes is crucial for reducing both random and systemic errors in RT-qPCR.The hallmark of a reliable reference gene lies in its stability across varied conditions to ensure accurate measurement of true expression disparities between samples [28].We selected six classical candidates from publications and previous experience.These candidates underwent RT-qPCR and were analyzed for stability across 0, 0.1, 1, and 10 µg/ml doses of MDP.The stability of the reference genes was assessed using (Supplementary Figure 1A and 1B).Additionally, we determined the optimal number of reference genes for normalization in our experiments (Supplementary Figure 1C and 1D).

Cell viability assay (MTS assay)
The CellTiter 96® Aqueous One Solution Cell Proliferation Assay kit (MTS) (Promega; cat#G3580) was used to evaluate cell viability across treatments.Following each study, the culture media was replaced by fresh media and incubated for 3 h with CellTiter 96® Aqueous One Solution.The resulting formazan product was quantified at 490nm, which is directly proportional to the number of living cells in culture.

Enzyme-linked immunosorbent assay (ELISA)
TNF-α and CCL5 gene expression data were validated by quantifying the protein released into the cell culture media using ELISA.At 70% confluency cells were treated with 0.1 µg/ml, 1 µg/ml, and 10 µg/ml MDP in 6 biological replicates for 24 h.Cell culture media was harvested and stored at -80°C or used immediately for the ELISA protocol.The remaining cells were used to assess cell viability as described above.To evaluate TNF-α concentrations in the media, the BioLegend Mouse TNF-α ELISA MAX™ Deluxe Set (cat#430904) was used according to the manufacturer's protocol.CCL5 was quantified using R&D systems mouse CCL5/RANTES DuoSet ELISA (cat#DY478-05) according to the recommended protocol.The optical density of each well was immediately measured using a Multiskan™ FC Microplate Photometer (Thermo Scientific; cat# 51119000).Cell culture media was used undiluted with all conditions measured in 3 technical replicates.

NF-κB translocation assay
To assess the influence of MDP on the NF-κB pathway in microglia, the translocation of NF-κB subunit p65 (RelA) into the nucleus was measured by immunocytochemistry.Cells were either treated for 60min with 10 µg/ml MDP or pre-treated with 10 µM SB202190 for 1 h followed by a 1 h incubation with 10 µg/ml MDP.Cells were fixed in 4% PFA (Merck; cat# 1.00496) for 10 min.After fixation, cells were washed three times with PBS.For blocking and permeabilization cells were incubated in staining buffer (SB) (0.5% BSA (Sigma; cat# A7030) + 0.1% triton X-100 (Sigma; cat# T8787) in PBS) for 1 h at room temperature (RT).
Next, cells were incubated with primary antibody anti-NF-κB p65 (Cell Signaling;

Statistical analysis
Data was analyzed and visualized using Rstudio (version 2023.06.2.561).After assessing normal distribution, a general linear model (GLM) with gaussian distribution was employed.If a specific gene violated the assumptions of the gaussian distribution, we applied a gamma distribution (detailed in Supplementary Table 2).The GLM was chosen due to its suitability for analysing complex interactions between multiple factors, such as MDP concentrations and pathway inhibitors.Estimated marginal means were calculated within the GLM framework to ascertain the mean differences among the factor levels.To correct for multiple comparisons, p-values were adjusted using the Holm´s method.This post-hoc test was chosen for its robustness in controlling family-wise error rates and its suitability for the present dataset's experimental design.Nuclear translocation of NF-κB was analysed using the one-way ANOVA for multiple comparison testing and Tukey's posthoc test.Values were considered significant p < 0.05.The dharma function in R was used to check for outliers.No outliers were detected hence no datapoints were removed.A detailed table of statistical methods and p-values can be found in Supplementary Table 2.
The number of wells per condition is defined as "n"-value for the RT-qPCR experiments.
For the immunocytochemistry experiments "n" was defined as number of wells (see 2.10), whereas individual datapoints shown are the number of images taken per condition.

Cells
IMG cells display common macrophage but also microglia-specific markers to distinguish them from brain infiltrating macrophages [23][24][25].As a primary step we confirmed the expression of microglia-specific markers P2Ry12 and TMEM119 in IMG cells (Figure 1A).Iba1 and CD11b stain both macrophages and microglial cells and are co-expressed in our IMG cells (Figure 1B).
To characterise the regulatory role of the PGN fragment MDP in microglia, we assessed the time course of microglial cytokine transcription following MDP treatment in IMG cells.
Next, we determined the expression trajectory of cytokines TNF-α (Figure 1F) and IL-1β (Figure 1G) as well as chemokine C-C Motif Chemokine Ligand 5 (CCL5) (Figure 1H).TNF-α expression increased significantly compared to baseline (~2-fold; p<0.001; Figure 1F) after 90 min of treatment, continued to increase and peaked at 3 h (~2.5-fold; p<0.001), and returned to baseline levels at 24 h.IL-1β exhibited a steep increase 90 min after MDP exposure, rising from 3-fold to 7-fold at 3 h compared to 0 h (p<0.001; Figure 1G), before returning to baseline levels at 24 h.The chemokine CCL5 was the only gene showing a late onset following MDP exposure, with its expression levels increasing after 3 h (~1.5-fold; p<0.001; Figure 1H) of treatment that continued with a steep elevation up to 24 h (~5-fold; p<0.001; Figure 1H) compared to baseline.Given that most changes occurred at 3 h and/or 24 h, we chose to focus more on these specific time points in subsequent studies (Figure 5 and 6).

Late Time-Dependent Modulation of Gene Expression Following MDP Exposure in IMG Cells
After assessing the early treatment effects of MDP on IMG cells, we identified a peak in expression for most of the measured genes.However, CCL5 exhibited a steep elevation at 24 h (Figure 1H), which suggests a delayed response even to extended periods of MDP exposure.
To determine the plateau in CCL5 expression, IMG cells were exposed to 10 µg/ml MDP for 24 h, 48 h, and 72h.
CCL5 expression rises after prolonged MDP treatment and plateaus at 48 h to 72 h.Between 48 h and 72 h, no signification difference was detected (Figure 2E) with similar results being obtained for CCL3 expression.After a significant increase from 24 h to 48 h (~1.3-fold; p < 0.05; Figure 2B), the expression started to plateau at 72 h.Interestingly, surface marker CD16 showed a decrease after 24 h and continued a marginal decline through the entire 72 h MDP treatment (Figure 2C).As depicted in Figure 1D, TNF-α expression peaked after 3h MDP exposure and returned toward baseline after 24h.By measuring TNF-α expression at later time points, we can appreciate an oscillation in expression after exposure to MDP.After the drop at 24 h, TNF-α expression increased to similar levels compared to after 3 h MDP treatment (~2fold; p<0.001; Figure 2D).No second peak was observed for IL-1β expression (Figure 2A).The gradual increase of CCL5 and TNF-α RNA expression led to a stepwise increase of cytokines in cell culture media (Figure 2F and G).Our results indicate that the expression of cytokines and chemokines in response to MDP differs greatly depending on the exposure time.TNF-α expression oscillated, whereas IL-1β showed an early (3 h) expression peak and CCL5 plateaus at a late (72 h) timepoint.

Validation of Late MDP Effects in Primary Microglial Cell Culture
To validate our findings in IMG cells, long-term and short-term effects of MDP were assessed using primary microglial cultures.In the first experiment, primary microglial cells were exposed to 10 µg/ml for 3 h, 24 h, and 48 h.In line with our findings in IMG cells, a peak in the expression of MDP receptor Nod2 (~7-fold; p< 0.001; Figure 3A) and NF-κB2 (~3.5-fold; p< 0.001; Figure 3B) can be observed at 3 h.Surface marker CD16 displayed a decrease in expression at the same time point (~0.3-fold;p< 0.001; Figure 3C).Similarly, a decrease in CD16 can be seen in IMG cells at 3 h (Figure 1C).CD16 then gradually returns to baseline expression over the next 48 h (Figure 3C).In primary microglia, TNF-α shows a 7.5-fold increase after 3 h MDP treatment (p< 0.001; Figure 6D), which returns to baseline after 24 h and stays constant after 48 h.In IMG cells, MDP exposure increases TNF-α expression less than 3-fold after 3 h (Figure 1D).Moreover, MDP treatment leads to increased expression of IL-6 (Figure 3F) in primary microglial cultures.We did not observe an effect of MDP on IL-6 expression in IMG cells.
Similar expression patterns of chemokines CCL3 and CCL5 were observed (Figure 3G, H); however, MDP has a stronger effect on cytokine expression in primary microglia.A more than 50-fold increase of IL-1β expression can be found at 3 h (p< 0.001; Figure 3E) compared to baseline, whereas CCL3 shows a 1.5-fold elevation (p< 0.001; Figure 3H).Chemokine CCL5 and CCL3 expression return to baseline after 24 h MDP treatment (Figure 3G, H), whereas in IMG cells, chemokine expression keeps rising after 24 h and plateaus after 72 h (Figures 2A and 2B).
In summary primary microglia exhibited an overall stronger response to MDP which normalized rapidly.

Validation of Early MDP Effects in Primary Microglial Cell Culture
We hypothesised that primary microglia are more active than IMG cells and as such we chose to look in more detail at the earlier timepoints after 90 min, 3 h, and 6 h of MDP treatment.
Compared to IMG cells, primary microglia exhibited a fast response to MDP.Overall, there was a very similar reactivity to MDP of both cell types, with peak expression after 90 min to 3h, supporting the usage of IMG cells as an in vitro method to assess gene regulation by MDP.

Dose-Dependent Patterns of RNA and Protein Expression in IMG cells Following Prolonged (24 h) MDP Exposure
To confirm our hypothesis that an optimal range of PGN is necessary for normal brain function, we explored transcriptional responses to physiological levels (0.1 µg/ml, 1 µg/ml) compared to high (10 µg/ml) doses of MDP.The latter was included to represent elevated levels (above 1 µg/ml) observed systemically in patients with autoimmune disorders such systemic lupus erythematosus and rheumatoid arthritis [27], conditions known to be associated with neuroinflammation and neuropsychiatric symptoms [30,31].Additionally, we sought to a potential threshold concentration of MDP that is required for microglia activation.
IMG cells were exposed to 0, 0.1, 1 or 10 µg/ml of MDP, and changes in gene (Nod2, NF-κB2, TNF-α, IL-1β, IL-6, and CCL5) and protein (TNF-α and CCL5) expression were evaluated 24 h after treatment.A dose-dependent effect was observed for most of the genes analysed.One of the clearest examples was CCL5 gene expression (Figure 5A) that significantly increased with each increasing MDP concentration.Both higher doses resulted in an upregulation of the NOD2 receptor and NF-κB2, with the latter showing a more robust effect (Figure 5B and C).
When examining cytokine and chemokine regulation, we observed approximately a 1.5-fold increase in TNF-α (p < 0.001, Figure 5D) and a roughly 2-fold increase in IL-1β (p < 0.01, Figure 5E) at 1 µg/ml compared to control.Interestingly, the expression of IL-6 was not affected by any dose of MDP (Figure 5F).Conversely, MDP induced a dose-dependent increase in CCL5 with a 5-fold upregulation after 10 µg/ml MDP challenge (Figure 5A).Notably, the lowest dose of MDP (0.1 µg/ml) had no significant effect on any of the evaluated genes, except CCL5.Thus, CCL5 appears to be the most sensitive to low MDP concentrations (Figure 5A).
To investigate if the changes in RNA translated into protein levels, we measured the amount of TNF-α and CCL5 protein secretion in IMG cell media following exposure to the same doses of MDP.TNF-α protein expression mirrored the changes observed in RNA expression (Figure 5G).Treatment with both 1 and 10 µg/ml resulted in a significant upregulation (p<0.001) of TNF-α compared to control, while expression was not significantly altered with 0.1 µg/ml (Figure 5G).However, there was a notable difference in CCL5 protein levels compared to the RNA expression profile in response to MDP exposure.Although we did observe significant increases in CCL5 protein levels at both 1 and 10 µg/ml compared to control (p< 0.001, Figure 5H), no differences in CCL5 protein were observed between the high doses of MDP.
Furthermore, protein expression did not match the significant increase in RNA expression at 0.1 µg/ml MDP.Cell viability was assessed before conducting protein quantification and no significant differences were recorded (Figure 3I).
In summary, a clear difference in the effect of low compared to upper physiological MDP doses on microglial activation was evident.Only chemokine CCL5 expression was increased after exposure to the lowest physiological MDP dose.In contrast, exposure to upper physiological levels of MDP resulted in expression of the pro-inflammatory cytokines TNF-α and IL-1β, but not of IL-6.Changes in TNF-α and CCL5 levels are reflected in protein levels in cell culture media after a 24 h exposure to MDP.

Short-term MDP Treatment Results in Nuclear NF-κB Translocation in IMG cells
As described above, MDP-induced activation results in long-term changes in gene expression as well as an increase in secretion of TNF-α and CCL5.As our initial investigations already revealed transcriptional changes at an earlier time-point, we chose to investigate the expression profile of genes specific to a short-term (3 h) exposure, which helped to complete the characterisation of microglial response to MDP.We hypothesized that microglia would respond differently to low vs high doses following short exposure to MDP.
Compared to long-term exposure (24 h), short-term exposure (3 h) to MDP resulted in a more pronounced concentration curve of most of the genes.Notably, Nod2 (Figure 6A), NF-κB2 (Figure 6B), and the cytokines TNF-α (Figure 6C) and IL-1β (Figure 6D) exhibited an early high expression, and significant differences between 1 µg/ml and 10 µg/ml MDP.Changes in CCL5 expression were observable at 3 h; however, the gene expression level at 3 h (1.3-fold, p<0.001, Figure 6E) was lower compared to 24 h (5-fold, p<0.001, Figure 6E).Time-dependent effects of MDP were clearly reflected in CCL3 expression, with only a significant increase after 3 h and not 24 h MDP treatment (see Figure 6F).
We next hypothesized that NF-κB translocation is one a of the main driving forces underlying the low-grade microglia response induced by MDP exposure.NF-κB is a key pathway activated for immune response initiation in microglia and is responsible for the transcription of a large family of inflammatory genes and subsequent production of related proteins [9].To assess the influence of MDP on the NF-κB pathway in microglia, the translocation of NF-κB subunit p65 into the nucleus was measured using fluorescence microscopy after 15 min and 60 min.
Compared to control, MDP exposure resulted in a 1.2-fold (p<0.05)increase in NF-κB signal after 15 min, and a 1.3-fold (p<0.001) increase after 60 min in the nucleus.In the control group, representative images revealed a clear segregation between the nucleus (stained with DAPI) and prominent NF-κB expression in the cytosol.An increase in NF-κB signal colocalizing with nuclear DAPI signal was observed after 15 and 60 min MDP treatment (Figure 6H).This indicated rapid regulation of NF-κB activation and translocation to the nucleus in microglia after MDP exposure and a mechanism for inflammatory protein expression.

MDP Signals via an Interplay of MAPK and NF-κB
Parallel to the NF-κB pathway, stimulation of Nod2 can result in the activation of mitogenactivated protein kinases (MAPKs) p38, ERK1/ERK2 and c-Jun N-terminal kinase (JNK) activation [32].To assess the involvement of MAPK p38 in downstream MDP signalling, IMG cells were treated with the p38 inhibitor SB202190 (SB20) prior to MDP exposure.SB20 blocked the expression of CCL5 after MDP exposure (Figure 7A).The MDP-induced increase in IL-1β expression (~5-fold; p<0.001; Figure 7B) was not only blocked but the expression was significantly decreased by SB20 compared to MDP treatment alone (p<0.01; Figure 7B).Treatment with SB20 resulted in a comparable IL-1β downregulation with or without MDP co-stimulation (see Figure 7B).Blocking the MAPK pathway thus eliminated the expression of IL-1β, with or without MDP.SB20 effectively attenuated the increase in NF-κB2 induced by MDP.While NF-κB2 expression was only slightly decreased upon SB20 exposure and comparable to control levels, MDP was not able to affect gene expression when the MAPK pathway was inhibited (Figure 7C).MDP significantly elevated TNF-α gene expression.
However, a reduction in MDP-induced TNF-α levels could be observed after SB20 treatment (Figure 7D), supporting a MAPK-mediated mechanism of action following MDP exposure in microglia.
We next hypothesized that MAPK plays a role in MDP-induced cytokine expression by influencing NF-κB translocation to the nucleus.We observed a significant reduction of NF-κB p65 signal in the nucleus between MDP treated and MDP+SB20 treated wells (p<0.001; Figure 7E).The increased NF-κB p65 translocation after MDP exposure was not only attenuated but slightly decreased by blocking p38 (Figure 7E).Representative images of IMG cells after MDP exposure reveal a translocation of NF-κB p65 into the nucleus, whereas SB20 and SB20+MDP resulted in NF-κB p65 primarily located in the cytosol similar to the control image.
The MAPK inhibitor eliminates IL-1β expression with and without MDP exposure, while also blocking MDP-induced expression of Nod2, NF-κB2 and CCL5.MAPK p38 inhibition seemed to be involved in TNF-α downregulation.As visualized in Figure 7G, translocation of the NF-κB subunit p65 after MDP treatment can be attenuated by blocking the MAPK pathway, leading to inhibition of gene expression.We therefore propose that the interplay of both MAPK and NF-κB pathways is important for MDP signalling in microglia.
Gut microbiota-brain communication is a rather new but rapidly emerging research field.
Multiple pathways have been identified through which signals from the gut can be conveyed to the brain (e.g. the vagus nerve and bacterial metabolites).Recent studies propose an alternative pathway via bacterial PGN fragments from the commensal microbiota, which can translocate from the intestinal mucosa into the circulation and reach the brain under physiological conditions [12,17,20].This study explores the impact of low-to-upper physiological, as well as high doses of MDP, the minimal bioactive peptidoglycan motifs common to all commensal bacteria, on microglia immune-related functions.MDP induces a time-and concentration-dependent pattern of cytokine and chemokine expression via Nod2 signalling.Furthermore, our results indicate that MDP-mediated microglia activation involves both the MAPK p38 and NF-κB pathways.Finally, similar time-dependent expression patterns in response to MDP were observed in primary microglial cell cultures, thereby validating our findings in IMG cells.
During the first years of postnatal life, major events in the brain including synaptogenesis, maturation and expansion of glial cells and myelination coincide with the development of the infant gut microbiota.The communication of the two distinct systems is crucial to ensure proper functioning.For example, bacterial-derived molecules from the gut microbiota have been associated with a typical development of neuronal circuits and microglia [2,33,34].
Moreover, germ-free (GF) mice have increased motor activity and decreased anxiety-like behaviour compared to mice with normal gut bacteria.Their behavioural phenotype was linked to altered second messenger pathways and synaptic transmission [35] Bacterial PGN, as a major component of the bacterial wall, has been investigated in several studies related to infection, and its infection-enhancing effects by priming human monocytes [36,37].Similar effects can be achieved by oral MDP treatment in mice, suggesting MDP released by gut bacteria may contribute to the priming process [38].When mice were gavaged with radiolabeled PGN ([ 3 H]-PGN), PGN could be detected in the brain as early as 2 h later.As this represents the earliest time point measured, the effects of MDP may manifest even sooner than 2 hours.Interestingly, the doses measured in the brain exceeded [ 3 H]-PGN concentrations measured in serum and other organs [17,20].
The exact mechanism of PGN translocation to the brain remains be elucidated.SLC15A family proton-coupled oligopeptide transporter proteins are the major transporters of PGN, with expression observed within the brain [39,40], but their expression has not been detected in the blood brain barrier (BBB).Since [ 3 H]-PGN can only be measured after gavage and not after intravenous administration, we suggest a membrane vesicle-based transportation mechanism or uptake by macrophages, dendritic cells and neutrophils [16,17].Interestingly, fluorescently labeled PGN was mainly absorbed by goblet cells in the intestines.Goblet cells are commonly known for immune surveillance but can potentially traffic PGN out of the gut lumen [17].MDP effects on peripheral cell types (e.g.monocytes [41]) have been reported previously, but knowledge about MDP's impact on cell types in the CNS is still lacking.Since the MDP receptor Nod2 is expressed in several brain regions but has its greatest expression in microglia in mice ( https://brainrnaseq.org/ ) [37], and our findings indicate a direct response of IMG cells and primary microglia to MDP, we propose that microglia play a role in PGN signalling in the brain.Our study therefore sheds light on the effect of the PGN fragment MDP on microglial function, and the MDP-mediated molecular mechanism of microglial activation.
We observed that MDP induces transcriptional changes in a large number of genes involved in Nod2 signalling, as well as cytokines and chemokines.In IMG cells a single dose of MDP can result in changes that lasted up to 72 h.MDP receptor Nod2 and transcription factor NF-B2 were significantly increased after 3 h, as well as pro-inflammatory cytokines TNF-α and IL-1β.
CCL5 emerged as a quite unique gene, as it showed a very late elevation in expression as response to MDP stimulation.We next extended our findings by looking at gene expressions up to 72 hours.After extended MDP exposure a normalization of IL-1β expression was observed while CCL5 and TNF-α expression stayed elevated.We could also verify that both CCL5 and TNF-α protein secretion was increased following MDP treatment.It is furthermore noteworthy that TNF-α exhibits an oscillating behaviour over time with a peak after 3 h and 72 h.The surface marker CD16 was used to monitor microglia activation, which is commonly expressed during phagocytic events and LPS stimulation [29].As we see an increase in proinflammatory cytokine expression, CD16 is the only gene decreasing in expression over time.
This shows that MDP does not lead to a strictly active/pro-inflammatory state but a complex phenotype with pro-inflammatory cytokine release and reduced expression of activation markers.
To confirm the validity of our findings we next assessed the changes in transcription following MDP treatment of primary microglia, which is a crucial step towards replicating our findings in the mouse brain.We could reproduce the findings from IMG cells with some notable differences.The effects in primary microglia occur more rapidly but normalize much faster, within 24 h.To expand the understating of the earliest time points we performed an additional experiment and found that most genes increased expression as early as 90 min, with a peak expression after 3 h of MDP treatment.In primary microglia the TNF-α expression profile did not only differ from the pattern observed in IMG cells but also from all other genes tested.
TNF-α had a maximum expression already after 90 min MDP treatment, which precedes the peak in expression of members of the Nod2 signalling pathway.This suggests that MDP regulates the expression of pro-inflammatory cytokines TNF-α and IL-1β via different downstream mechanisms.Overall, the changes in gene expression were stronger for all the proinflammatory genes and members of the signalling pathway in primary microglia vs IMG cells, while chemokines exhibited a weaker expression.
Due to the strong correlation between findings in IMG cells and in primary glia we continued with the follow-up studies using IMG cells.In 75% of healthy individuals from European or Asian ancestry, PGN serum levels can range between 0.16-1 μg/ml, whereas 22.5% lie between 1-4 μg/ml and few over 4 µg/ml [19].Our next step was therefore to assess the expression of our key genes identified earlier and expose them to different concentrations of MDP.We determined that the lowest physiologically prevalent dose (0.1 µg/ml) of MDP increased the expression of CCL5, whereas the expression of pro-inflammatory cytokines TNFα and IL-1β was only affected at higher concentrations.We observed an increase in TNF-α and IL-1β at upper physiological (1 µg/ml) to high (10 µg/ml) levels of MDP, indicating that 1 µg/ml potentially represents a threshold for MDP-induced neuroinflammation.For instance, patients with autoimmune disorders such as systemic lupus erythematosus and rheumatoid arthritis have increased PGN serum levels (> 1 µg/ml) [27], as well as a predisposition in developing neuropsychiatric symptoms [31].
Elevated levels of MDP have been recorded in the serum of patients with schizophrenia compared to control individuals [41], suggesting PGN levels outside of the normal range can contribute to disease.While PGN was detected in the white matter of multiple sclerosis patients and antibodies against PGN were found in their cerebral spinal fluid [16,42], we cannot extrapolate serum levels to the brain.We therefore cannot directly translate concentrations observed in patients and healthy individuals to our findings in vitro.High levels of MDP in the brain might be associated with impairment of the BBB function.Previous studies have reported an increased permeability of the BBB in a schizophrenia mouse model, as well as in schizophrenia patients, supporting this hypothesis [43,44].Optimal levels of MDP might be furthermore required for a typical brain development, particularly during crucial developmental stages such as microglial maturation and synaptogenesis [19].Early detection of abnormal PGN levels and interventions during pregnancy could therefore help avoid neurodevelopmental disease outcomes in offspring.
Interestingly, the expression of IL-6, a common pro-inflammatory cytokine, remained unchanged following MDP exposure in IMG cells and was transiently and much less affected in primary microglia compared to the other pro-inflammatory cytokines, indicating an MDPspecific cytokine expression pattern rather than an overall expression of pro-inflammatory cytokines.We previously reported a strong pro-inflammatory effect of bacterial LPS from the wild-type S-form E. coli serotype O55:B5, when challenging IMG cells in culture [25].LPSinduced gene expression of the major pro-inflammatory cytokines included IL-6.We found 60fold higher TNF-α protein levels after LPS challenge compared to our findings after MDP exposure [25].In this current study MDP exposure leads to a low-grade microglial activation compared to the pro-inflammatory microglial phenotype induced by LPS challenge.The MDPinduced secretome might stimulate further processes such as neurodevelopment and plasticity in the brain.We identified the chemokine CCL5 as a key target of low, physiologically relevant doses of MDP.CCL5 plays an important role in memory formation and synaptic plasticity in vivo [45].Interestingly, intraperitoneal injections of MDP without co-stimulant have been shown to upregulate the expression of postsynaptic density protein 95 (PSD-95) and monocyte chemoattractant protein-1 (MCP-1) in the cortex and hippocampus [46].CCL5 released by microglia therefore could be a potential effector of MDP-dependent alterations in synaptic gene expression.
We also demonstrated MDP-induced NF-κB signalling by performing an NF-κB nuclear translocation assay.The translocation of NF-κB subunit p65 to the nucleus increased after MDP treatment compared to the control.Notably, we found that MDP-induced activation of the NF-κB pathway is regulated by MAPK p38.The upregulation of NF-κB2, as well as IL-1β and CCL5 upon MDP treatment, is either normalised or blocked by MAPK p38 inhibitor treatment.
Previous studies have reported a p38 dependency of NF-κB p65 activation in astrocytes [47], and our findings are consistent with these observations.Therefore, the interplay between the MAPK and the NF-κB pathways might be responsible for the microglial phenotype during exposure of IMG cells to non-pathogenic MDP levels.
Interestingly, Jeric and colleagues (2023; [48]) reported a specific modification of PGN, anhydro-PGN, from a probiotic bacterial strain that exhibited anti-inflammatory effects in macrophages.Anhydro-PGN reduced LPS-induced cytokine expression through an unknown alternative pathway to the NOD signalling pathways [48].Together with our results using commensal MDP, those findings reflect the diverse effects of PGN on the host.Additionally, a new group of neuroactive BBB-penetrant microbial metabolites has been reported, which directly affect neuronal circuits and behaviour through altered myelination [49].Those metabolites were identified in the plasma and faeces of patients with autism spectrum disorder (ASD), indicating new potential therapeutic approaches [50].Taken together our study and recent findings underscore the importance of studying bacterial metabolites and components from the commensal gut microbiota and the underlying mechanisms in the context of typical brain development to identify further therapeutical targets.
Our results represent just the first step into a newly evolving field.Working with cell lines yields possible limitations compared to in vivo experimentation.The brain harbors a complex environment that cannot fully be mimicked in vitro.However, it has already been established that bacterial PGN impacts the development of the brain and behaviour in mice [19].Our study aimed to provide insight into the mechanism underlying the regulatory role of PGN in the microglial niche of the brain.More research is necessary to investigate the effect of MDP from different bacterial strains (probiotic and pathogenic), as well as further characterization of potential alternative pathways.The ability to use the IMG cell line significantly speeds up studies, replaces the use of animals and enables studies with more power to be performed.

Conclusions
In conclusion, our study demonstrates the direct effect of MDP on IMG and primary microglial cells.MDP levels of 1 µg/ml and higher increase the expression of pro-inflammatory cytokines by engaging the NF-κB and MAPK pathways.MDP may thus serve as a mediator for inflammatory processes in the brain.These findings emphasize the importance of low compared to high physiological levels of PGN in shaping microglia function in the CNS.

Supplementary table 2.4 Statistics cytokine expression (RT-qPCR)
cat#6956S; dilution 1:500; mouse) in SB overnight at 4°C.The next day, cells were washed 3 times in PBS (5min per wash) and incubated with secondary antibodies goat anti-mouse 555 (Abcam; cat# ab15001; dilution 1:800) and DAPI (invitrogen; cat# D3571; dilution 1:5000) in SB for 1 h shaking at RT in the dark.After incubation with the secondary antibody, cells were washed 3 times in PBS (5 min per wash) and stored in PBS at 4°C.The NF-κB signal was measured using fluorescence microscopy.Images were analyzed using Fiji to compare for single-cell nuclear NF-κB signals (5 wells per condition; MAPK inhibitor study: total cells per condition > 15000, MDP time curve > 1500).The following workflow was used in batch mode: Separate channels > Maximum intensity projection > Threshold (DAPI) > Watershed (DAPI) > Analyse particles (DAPI) > Select NF-κB channel > Overlay from ROI Manager > Measure (NF-κB).NF-κB signal was normalised by DAPI signal.