Animals

All experiments were performed in accordance with the ARRIVE guidelines 2.0 (Additional file 1: Table S1) [32], with the German legislation on protection of animals and with approval of the local animal welfare committee (Thüringer Landesamt für Lebensmittelsicherheit und Verbraucherschutz, UKJ-18-026). A total of 7 experiments were performed using 193 male C57BL/6J mice, aged 10–13 weeks, from the in-house breeding facility. Animals were randomized to the treatment groups, with equal distribution of age and weight in all four groups. Animals were housed under standardized day-night conditions (12 h/12 h) at room temperature (23 ± 1 °C, 30–60% humidity). Mice were fed with AIN-76A (Research Diets, Inc.) standard chow and water ad libitum.

Peritoneal contamination and infection (PCI) sepsis model and PLX5622 treatment

Polymicrobial sepsis was induced by the standardized peritoneal contamination and infection model (PCI) and sterile systemic inflammation using lipopolysaccharide (LPS) administration [33, 34]. Human fecal slurry in the PCI model (diluted 1:4 in saline solution; 3.5 µl/g B.W.) or LPS (3 mg/kg B.W.; Sigma-Aldrich, Germany) were injected into the left lower quadrant of the abdomen with a 21-gauge cannula. To increase survival rates and ensure comparability, sham and PCI treated animals received subsequent antibiotic treatment. Therefore, meropenem (20 mg/kg B.W.) was injected s.c. every 12 h for 7 days starting at a clinical severity score (CSS) of 3, indicating severe infection, in both sham and PCI mice [33]. Between day 8–10 meropenem was applied once daily. Thereafter, enrofloxacin (Baytril 2.5%, Bayer AG, Germany) was diluted in saccharose containing drinking water (final concentration of 2 mg/ml) and applied until the end of the experiment to avoid re-occuring infection [23]. Disease severity was continuously assessed using the Clinical Severity Score (CSS; grade 1–4) [33]. Grade 1: No signs of illness, active and curious, quick movements upon exogenous stimuli, normal posture; grade 2: Low-grade illness, less active with occasional interruptions in activity, reduced alertness, but adequate response to exogenous stimuli, posture slightly hunched; grade 3: moderately severe illness, slow and sleepy, movement difficulty, limited and delayed reaction to exogenous stimuli, hunched posture; grade 4: severe illness, mouse lethargic, motionless, no spontaneous movements, no reaction to exogenous stimuli, severely hunched posture. Floating numbers between two grades are possible. For ethical reasons, mice were euthanized if a CSS of grade 4 was reached at two consecutive time points within 3 h and mice were not included in further analysis. To ensure adequate and comparable disease severity between groups, only mice with a 3-day cumulative CSS greater than 9.5 and a 5-day cumulative CSS greater than 10.5 were included in the study for further analysis.

PLX5622 was provided by Plexxikon Inc. and is formulated in AIN-76A standard chow in given dosages of 1200 mg/kg and 300 mg/kg. PLX5622 needs to be administrered for at least 7 days to achieve stable microglia depletion [26, 28]. High-dose PLX5622 (1200 ppm; 1200 mg/kg; Plexxicon Inc. Berkeley, CA, USA) or low-dose PLX5622 (300 ppm; 300 mg/kg; Plexxicon Inc. Berkeley, CA, USA) containing AIN-76A (AIN) standard chow in the microglia depletion groups was started seven days prior to PCI or LPS induction in comparision to control diet AIN. At the time point of PCI induction or LPS administration PLX5622 treatment was ended. Mice were continuously weighed before and after sepsis induction or LPS administration to assess adequate uptake of PLX5622 and to evaluate disease severity during sepsis.

Immunohistochemistry

At the end of experiments, animals were sacrificed by overdosage of isoflurane (2%) and were transcardially perfused with phosphate buffered saline (PBS). Brain tissue was harvested for further ex-vivo experiments at the respective time points. After preparation, the dissected brains were cut in half and one hemisphere was fixated for 24 h in 4% paraformaldehyde (PFA), dehydrated for 24 h in 10% and 24 h in 30% sucrose solution. The PFA-fixed hemispheres were stored at − 80 °C. Coronal sections were prepared with a cutting microtome (16 or 40 µm) (Thermo Scientific Microm HM 450, Thermo Fischer Scientific, Schwerte, DE, USA) and stored at − 20 °C as free-floating sections in antifreeze solution.

For immunohistochemistry, slices were washed with TRIS buffered saline (TBS) and blocked with blocking buffer (3% serum, 2% milk powder and 0.1% Triton X-100 in TBS) for 30 min. Thereafter, slices were incubated overnight at 4 °C with the primary antibody against the ionized calcium-binding adapter molecule 1 (Iba1) (rabbit, 1:500, Cat# 019-19741, RRID:AB_839504, FUJIFILM Wako Shibayagi), the glial fibrillary acidic protein (GFAP) (mouse, 1:500, Cat# MAB360, RRID:AB_11212597, Millipore) or Homer1 (chicken, 1:500, Cat# 160006, RRID:AB_2631222, Synaptic Systems). Following washing steps in TBS, slices were incubated for 2 h in the secondary antibody (AF488, donkey@rabbit, 1:500, Cat# A-21206, RRID:AB_2535792, Molecular Probes; AF488, 1:200, goat@chicken, Cat# A-11039, RRID:AB_142924, Thermo Fisher Scientific; AF647, 1:500, donkey@ms, Cat# 715-605-151, RRID:AB_2340863, Jackson ImmunoResearch Labs; AF647, 1:500, goat@rabbit, Cat# A-21244, RRID:AB_2535812, Thermo Fisher Scientific). Following washing in TBS, slices were transferred to microscope slides with 0.5% gelatin, dried and stained for 5 min in 4′,6-Diamidino-2-phenylindol (DAPI) (Sigma Aldrich, Cat# D9542, CAS-Nr. 28718-90-3, 1 mg/ml) solution and washed in PBS. Cover slips were mounted with Fluoromount G (Southern Biotech, Birmingham, AL, USA). Immunostained brain slices were imaged with a confocal laser scanning microscope (Zeiss LSM 900). For colocalization analysis regarding the synaptic engulfment assay the Airyscan mode of the microscope providing high-resolution images was used.

Pharmacological microglia depletion and microgliosis

For quantitative microglia analysis, Iba1 positive cells were counted in the CA1 region of the hippocampus. Therefore, murine brain sections (40 μm section thickness) were subsequently stained with commercial antibodies against Iba1 and DAPI. Representative Z-stack images were obtained (20 × objective, NA = 0.8). Z-stacks were scanned at 2048 × 2048 resolution, with 4 pixel-wise plane scan intensity averages for each color channel, and 1 μm step size to generate confocal stacks of ~ 5 μm per image. Finally, maximum intensity projections were generated and Iba1 microglia were manually counted in Fiji ImageJ 2.3.0 [27]. Only Iba1 positive cells with a DAPI-positive nucleus were counted. For the overview images tile scans (20 × objective, NA = 0.8) were generated. Cell count was normalized to the average cell count of the control condition AIN-sham. Experimenters were blinded to the group allocation.

GFAP-staining and quantitative analysis

For astrocyte analysis, slices were stained for GFAP and DAPI. To quantify hippocampal astrocytes (20 × objective, NA = 0.8) z-stacks were taken in the CA1 region (d3: n = 5 animals/group, n = 3 images/animal; d38: n = 5 animals/group, n = 2 images/animal). A standardized z-stack thickness of 20 µm with a z-interval of 1 µm per plane was used for each color channel. Cells were manually quantified using Fiji. Only GFAP-positive cells with a DAPI-positive nucleus in the center plane at 3 µm of the image stack were counted. Cell count was normalized to the average cell count of the control condition AIN-sham. Experimenters were blinded to the group allocation.

Quantitative analysis of engulfed synapses

For the colocalization analysis of synaptic engulfment, slices were stained for Iba1, Homer1 and DAPI. Representative z-stack images (d3: n = 4–5/group, n = 2–4 microglia/animal; d38: n = 5/group, n = 2microglia/animal) were acquired in the CA1 region of the hippocampus (63 × oil objective, NA = 1.4, Airyscan). Z-stacks were acquired with a 1490 × 1490 image size and with 4 pixel-wise plane scan intensity averages for each color channel and 16.66 µm z-step size. After image acquisition, "surfaces" were created in Imaris (version 9.1.2.; Oxford instruments) for Iba1 and Homer1. Using the "colocalization" function and an implemented MatLab code in Imaris, a fluorescence colocalization of Iba1 and Homer1 was performed. In the following, a mean threshold filter of 0.1 µm3 was applied to the colocalized volumes, based on the mean volumes of the colocalized volumes of both groups.

Open field test

Exploratory activity and anxiety-like behavior following sepsis induction were measured by open field test. The open field test was performed on day 9 and 29 after PCI induction one day prior to each novel object recognition (NOR) test. The experiment was conducted in a black open-field box (40 cm), illuminated by 15 lx LEDs, by an experimenter blinded to the group allocation of the two intervention strategies—PCI injection and PLX5622 treatment. Mice were placed in the open field box for 10 min (5 min for day 29) and were able to discover the box freely. The mean velocity and total distance were measured automatically on each day using the tracking software EthoVision (Ethovision XT® 6.1 software (Noldus Wageningen Netherlands) [35]. Between each trial, the box was carefully cleaned with 70% ethanol to ensure equal experimental conditions.

Novel object recognition test

To test learning and memory after sepsis, novel object recognition (NOR) test was performed as described previously [23]. NOR was performed at day 10 and 30 after PCI induction. The experiment was conducted in a black open-field box (40 cm), illuminated by 15 lx LEDs, by an experimenter blinded to the group allocation of the two intervention strategies—PCI injection and PLX5622 treatment. Objects were distinguishable and different for each testing day: an iron bracket and an objective container (both day 10), a small plastic cylinder, and an iron eyebolt (both day 30). The objects were previously tested and no preferences for an individual object could be found. Additionally, each object was randomized to be used as familiar or novel object. Between each trial, the box was carefully cleaned with 70% ethanol to ensure equal experimental conditions.

During the familiarization phase, mice were allowed to explore two equal objects for max. 10 min. The familiarization phase was terminated by the investigator after mice had explored the two objects for cumulative 20 s or 10 min if a cumulative exploration time of 20 s was not reached. In the testing phase, 6 h after the training phase, one object was replaced by the complementary novel object, and mice were allowed to explore the familiar and new object. The position of the object, left vs. right, was also randomized between animals during the test phase. Exploration time was measured and the session was completed after 20 s cumulative exploration behavior. Mice, which did not explore the objects for 20 s within 10 min in the familiarization or testing phase were excluded from the analysis. Every touching of the object and direct facing an object ≤ 2 cm apart from the object was considered as exploration behavior. Climbing onto the object was not regarded as exploration behavior.

Statistics

Researchers performing the analyses were blinded to the experimental treatment of the animals. Based on our previous results with the PCI mouse model [23], we performed a statistical power analysis to predict the minimum sample size per treatment group which was found to be close to 6 (p = 0.05 and 80% power). All graphs are represented as super-plots showing means ± standard error of the mean (SEM). Individual values are presented as small dots and each circle represents an average of one mouse. The latter was used for statistical analysis. Analysis was performed using R (1.4.1717) und R Studio (2021.09.0), OriginPro (v. 2020b), SigmaPlot (v. 14.5), and Matlab 2020a. If data were normally distributed (Shapiro–Wilk test), two-tailed unpaired Student’s t-test (2 groups) or Two-way ANOVA (4 groups, post hoc: Tukey test) were used. If data were not normally distributed, Mann–Whitney-U-Test (2 groups) were used. Data were also tested for homogeneity of variance (Levene test) and outliers (Grubb’s test). For Kaplan–Meier analysis log-rank test (post-hoc: Benjamini–Hochberg correction) was performed. Permutation tests were performed based on ref. [36] as follows: to compare the differences in the CSS, we adapted the two-tailed permutation test of Cohen (significance level of 5%), thereby accounting for multiple comparison problem. For the weight development between d-7-d0 and between d0-d38, we performed the two-tailed curve-permutation test. For days d1-d38, we subjected p-values to Benjamini–Hochberg procedure in order to account for multiple comparison problem. For each permutation test, 1 million times shuffling was used. Figures 1A, E, 2A, 6A, D were created with BioRender.com.

Fig. 1

High-dose PLX5622 pretreatment causes poor outcome in LPS and PCI induced systemic inflammation. A Timeline of experiment: mice were randomized to AIN or PLX5622 (1200 ppm) treatment seven days prior to PCI induction. B Quantitative analysis of Iba1-positive microglia in the CA1 region of the hippocampus at day 0 following AIN or PLX5622 treatment (n = 3 animals/group; n = 3 images/animal, Student’s t-test). Representative confocal overview images of microglia in the CA1 region after pretreatment with either AIN or high-dose PLX5622 (1200 ppm) on day 0. Scale bar 200 µm. C Kaplan–Meier analysis of PCI induced sepsis following AIN or high-dose PLX5622 treatment for seven days (AIN: PCI: n = 21 animals/group, PLX5622: PCI: n = 6 animals/group, log-rank test). D Representative photograph of cerebral microbleeds in mice pretreated with high-dose PLX5622 for seven days followed by PCI induction. E Timeline of experiment: mice were randomized to AIN or high-dose PLX5622 (1200 ppm) treatment seven days prior to induction of systemic inflammation by LPS. F Kaplan–Meier analysis of LPS induced systemic inflammation following AIN and high-dose PLX5622 pretreatment for 7 consecutive days (AIN: LPS: n = 14 animals/group and PLX5622: LPS: n = 6 animals/group). Data are presented as mean ± SEM. Individual values are presented as small dots and each circle represents an average of one mouse; individual values and averages are color coded

Fig. 2

Partial microlgia depletion by low-dose PLX5622 only slightly affects the course of sepsis. A Timeline of experiment: mice were randomized to AIN or low-dose PLX5622 (300 ppm) treatment seven days prior to PCI. After PCI (day 0) all mice received standard AIN chow. B Quantitative analysis of Iba1-positive microglia in the CA1 region of the hippocampus following AIN or PLX5622 treatment at day 0 (n = 3 animals/group; n = 15 images/animal, Student’s t-test). Representative confocal overview and zoom-in images of Iba1-positive microglia in the CA1 region after pretreatment with either AIN or PLX5622 (300 ppm). Scale bar 20 µm and 200 µm. C Kaplan–Meier analysis of PCI induced systemic inflammation following AIN and low-dose PLX5622 between day 0 until day 38 (AIN-sham: n = 20, AIN-PCI: n = 42; PLX5622-sham: n = 20, PLX5622-PCI: n = 45, log-rank test: p < 0.0001). D Clinical severity score (AIN-sham: n = 20, AIN-PCI: n = 42; PLX5622-sham: n = 20, PLX5622-PCI: n = 45; the adapted two-tailed permutation test of Cohen (significance level of 5%) thereby accounting for the multiple-comparison problem has been performed. E Time dynamics in the development of clinical severity score until CSS ≥ 3 (AIN-PCI: n = 42, PLX5622-PCI: n = 45, Mann–Whitney-U-test). F Weight development during pretreatment of AIN or low-dose PLX5622 between day -7 until day 0 and during polymicrobial sepsis between day 0 and day 38 (AIN: sham: n = 20, PCI: n = 42; PLX5622: sham: n = 20, PCI: n = 45; day -7 until day 0: Curve-permutation test (two-tailed), p = 0.3780; day 0 until day 38: Curve-permutation test (two-tailed) with Benjamini–Hochberg correction. Data are presented as mean ± SEM. Individual values are presented as small dots and each circle represents an average of one mouse; individual values and averages are color coded