Neural Injury and Repair in a Novel Neonatal Mouse Model of Listeria Monocytogenes Meningoencephalitis

Abstract

To improve the therapy of neonatal central nervous system infections, well-characterized animal models are urgently needed. The present study analyzes neuropathological alterations with particular focus on neural injury and repair in brains of neonatal mice with Listeria monocytogenes (LM) meningitis/meningoencephalitis using a novel nasal infection model. The hippocampal formation and frontal cortex of 14 neonatal mice with LM meningitis/meningoencephalitis and 14 uninfected controls were analyzed by histology, immunohistochemistry, and in situ tailing for morphological alterations. In the dentate gyrus of the hippocampal formation of mice with LM meningitis/meningoencephalitis, an increased density of apoptotic neurons visualized by in situ tailing (p = 0.04) and in situ tailing plus immunohistochemistry for activated Caspase-3 (p < 0.0001) was found. A decreased density of dividing cells stained with an anti-PCNA-antibody (p < 0.0001) and less neurogenesis visualized by anti-calretinin (p < 0.0001) and anti-calbindin (p = 0.01) antibodies were detected compared to uninfected controls. The density of microglia was higher in LM meningitis (p < 0.0001), while the density of astrocytes remained unchanged. Infiltrating monocytes and neutrophilic granulocytes likely contributed to tissue damage. In conclusion, in the brains of LM-infected mice a strong immune response was observed which led to neuronal apoptosis and an impaired neural regeneration. This model appears very suitable to study therapies against long-term sequelae of neonatal LM meningitis.

Infections of the central nervous system, Listeria monocytogenes, Neonatal mouse model, Neural injury and repair

INTRODUCTION

Bacterial meningitis is one of the most serious infections in neonates and is associated with a high mortality as well as a high rate of long-term sequelae (1, 2). Infants surviving meningitis have a significantly higher risk of developmental disabilities, particularly learning and neuromotor impairments (3). One of the 3 most important bacterial pathogens of neonatal meningitis is Listeria monocytogenes (LM) (4). Compared to other neuro-invasive organisms, LM is very efficient at invading the central nervous system (CNS) (5). Via the hematogenous route, LM reaches the CNS by crossing the blood-brain or blood-CSF barriers as a cargo of monocytes or by invasion of endothelial cells. CNS infection in neonates manifests as meningitis or meningoencephalitis (6, 7).

Perinatal infections of the fetus or the neonate with LM are divided into early-onset disease (EOD) and late-onset disease (LOD). The clinical pattern of each form is distinct. During EOD (≤1 week after birth) CNS invasion is thought to occur via the hematogenous route after placental infection. In 75% of these cases, blood cultures of newborns are positive for LM, and meningitis is seen only in 10% of the cases. In LOD the infection is established by contact of the neonate to the maternal microbiota of the vagina or gut during delivery. CNS infection is diagnosed in 95% of LOD cases, yet only 20% of blood cultures are positive for LM (8). The question thus arises which path LM takes to infect the CNS during LOD thereby bypassing the blood stream. In murine studies using parenteral application of LM, infection of the CNS is detected in only 25% of the cases (9).

Until recently, there was no infection model that reflected the tropism of LM for the CNS during LOD. In 2018, Pägelow et al (10) described a new infection model that demonstrated that nasal inoculation of newborn mice causes (i) colonization of the nasal mucosa, (ii) infection of the olfactory epithelium, (iii) transport of LM via axons of the olfactory sensory neurons through the cribriform plate and, finally, (iv) infection of the olfactory bulb and spreading of LM to other brain regions leading to meningitis or meningoencephalitis in all animals exposed to LM. This infection pathway could be important in the pathogenesis of LM LOD in human neonates since LM has been cultured from nasal swabs of LM-infected newborns (11).

CNS infections in newborns often affect cerebral maturation and neurodevelopment (12). In children surviving bacterial meningitis, disabilities can persist life-long (13–16). In rodent models, learning impairments caused by bacterial meningitis are connected to apoptosis of dentate granule cells in the hippocampal formation (17, 18). Well-characterized animal models that reproduce the pathogenesis of human listeriosis are urgently needed to develop therapies aimed at decreasing long-term sequelae especially in vulnerable individuals including neonates. The aim of this study was to characterize histopathological alterations in this novel intranasal infection model mimicking LOD with a special focus on neural injury and repair.

MATERIALS AND METHODS

Ethics Approval

Mice were cared for in accordance with the principles outlined in the European Convention for the Protection of Vertebrate Animals Used for Experimental and Other Scientific Purposes. The animal experiments were approved by the Committee on Animal Experiments.

Brain Samples

Brains of neonatal mice from partially published experiments (10) were analyzed. Briefly, 14 1-day-old neonates (C57BL/6) were intranasally infected with approximately 1 × 10⁴ colony forming units LM strain EGDe (WT 446) in 1 µl phosphate-buffered saline (PBS). Control mice received an equal volume of PBS. All mice were killed 5 days post-infection or 5 days after administration of PBS, and then autopsied according to the same protocol which included a predefined collection of samples for histological investigation. Brains were fixed with 4% paraformaldehyde for at least 24 hours, then dehydrated and embedded in paraffin wax.

Brain sections of the hippocampal formation and the frontal brain were used for further analysis by histology, immunohistochemistry, in situ tailing, and immunofluorescence. Infiltrating cells and brain tissue damage were evaluated by hematoxylin and eosin (H&E) staining of 1.5-µm sections from the frontal cortex and the hippocampal formation. H&E-stained sections were also used to detect ischemic lesions.

Immunohistochemistry, In Situ Tailing and Immunofluorescence Staining

1.5-µm brain sections of the hippocampal formation were deparaffinized, pre-treated with citrate buffer (10 mmol/L, pH 6.0), heated by microwaving (5 × 3 minutes; 800 W), blocked by incubation in 3% hydrogen peroxide (H₂O₂) for 10 minutes and then blocked with 10% fetal calf serum in PBS or Tris-buffered saline (TBS) for 30 minutes. All primary antibodies were applied at the concentrations indicated below and incubated overnight at 4°C in PBS or TBS. For the detection of proliferating cells, hippocampal sections were stained with a monoclonal mouse anti-PCNA (proliferating cell nuclear antigen) antibody (dilution 1:200; Chemicon, Temecula, CA). Microglia were detected using a polyclonal rabbit anti-Iba-1 (ionized calcium-binding adaptor molecule-1) antibody (1:400, Wako, Neuss, Germany). Astrocytes were visualized by a polyclonal rabbit anti-glial fibrillary acidic protein (GFAP) antibody (1:1000; Dako, Hamburg, Germany), and axonal injury was analyzed using a monoclonal mouse antibody against APP (amyloid precursor protein) (1:2000; Chemicon, Temecula, CA) as described (19). For staining of young post-mitotic neurons antibodies directed against Calretinin (1:1000, Swant, Bellinzona, Switzerland) and calbindin were used (1:150, Sigma-Aldrich, St. Louis, MO). For the detection of LM antigens, brain sections were blocked with 5% normal donkey serum in PBS for 1 hour at room temperature and then incubated with a polyclonal serum raised against LM in a rabbit (1:3000) overnight at 4°C. Secondary biotinylated monoclonal anti-mouse antibodies (Amersham, Buckinghamshire, UK) or biotinylated polyclonal anti-rabbit antibodies (GE Healthcare, Buckinghamshire, UK) were diluted 1:200 in PBS, followed by the addition of avidin–biotin peroxidase complex (1:1000, Dako), and diaminobenzidine (Roche, Mannheim, Germany) as chromogenic substrates. Binding of the calbindin antibody was visualized by the alkaline phosphatase/anti-alkaline phosphatase reaction developed with Fast Red (Sigma-Aldrich, Merck, Darmstadt, Germany) as the chromogenic substrate. In case of fluorescence staining, the secondary donkey anti-rabbit antibody was labeled with Cy3 and incubated on the slices for 1 hour at room temperature (dilution 1:1000) (Supplementary Data Table S1).

Brain sections were counterstained with hemalum (Merck) for 30 seconds at room temperature for immunohistochemical staining and counterstained with DAPI (1:1000, Invitrogen by Thermo Fisher Scientific, Dreieich, Germany) for 5 minutes at room temperature for immunofluorescence staining. Isotypic antibodies and brain slices incubated with secondary antibodies in the absence of primary antibodies served as controls (19). After incubation with primary and secondary antibodies, the samples were washed with PBS or TBS.

Apoptotic neurons in the dentate gyrus of the hippocampal formation were identified by in situ tailing combined with the immunohistochemical detection of activated caspase 3 by a polyclonal rabbit antibody (1:100, BD Pharmingen, Franklin Lakes, NJ) or morphological criteria (condensed, shrunken nuclei, condensed and shrunken eosinophilic cytoplasm). Briefly, deparaffinized brain sections were incubated with protein kinase K (Sigma-Aldrich, St. Louis, MO) for 15 minutes at 37°C. Thereafter, brain sections were incubated with the tailing mix (Sigma Aldrich, St. Louis, MO) for 1 hour at 37°C and incubated with anti-digoxigenin-antibody conjugated with alkaline phosphatase AP (1:250, Roche, Penzberg, Germany). The developmental substance was nitro-blue tetrazolium/5-bromo-4-chloro-3’-indolyphosphate (NBT/BCIP) (Sigma-Aldrich, St. Louis, MO). The primary polyclonal rabbit anti-activated Caspase 3 antibody (1:50, Dako) was incubated over night at 4°C on the brain slices. A monoclonal mouse anti-rabbit antibody (1:50, Dako) and a polyclonal rabbit anti-mouse antibody (1:50, Dako) were used to visualize early stages of apoptotic neurons. Fast red (Sigma-Aldrich, St. Louis, MO) was used as the developmental substance. The counterstaining was performed with hemalum for 30 seconds. Representative results of stained brain sections are shown in Figure 1.

FIGURE 1.

Histology, immunohistochemistry, and in situ tailing of brain sections from neonatal mice with LM meningitis/meningoencephalitis. H&E stainings of brain sections of mice suffering from meningitis (A) or meningoencephalitis (B). Apoptotic neurons were visualized by in situ tailing plus morphological criteria (dark violet nuclei, shrunken cytoplasm) (C) or in situ tailing plus staining for caspase 3 (dark violet nuclei with a pink cytoplasm) (D) in the hippocampal dentate gyrus. Dividing cells in the dentate gyrus were detected by staining of the proliferating cell nuclear antigen (PCNA) (brown) (E). Young neurons were stained with an anti-calretinin (brown) (F) and an anti-calbindin (pink) (G) antibody. Microglia were visualized by an anti-ionized calcium binding adaptor molecule 1 (Iba-1) antibody (brown) (H). Astrocytes were stained with an anti-glial fibrillary acidic protein (GFAP) antibody (brown) (I). Horizontal bars: A, B = 100 µm; C–I = 20 µm.

Microscopy and Statistical Analysis

To quantify immunoreactive cells in the dentate gyrus, only cells in the granule cell layer, the subgranular zone, and the hilus were counted. The Analysis Software Imaging System (microscope BX51; Olympus; software AnalySIS 3.2; Soft Imaging System GmbH, Münster, Germany) was used to measure the area of the dentate granule cell layer. For analysis of microglial activation, the morphological activation steps identified and described by Kreutzberg (20) were adapted (0 = fully resting microglia with multiple branches; 1 = mix between resting and mildly activated microglia, demonstrating shortened processes of increased thickness; 2 = mildly activated microglia with shortened processes of increased thickness; 3 = mix between mildly and strongly activated ameboid microglia without processes; 4 = ameboid microglia without processes) (Supplementary Data Fig. S1).

Data were expressed as medians with interquartile ranges visualized as scatter dot plots and compared for statistically significant differences by two-tailed Mann-Whitney U-test. Statistical analyses were carried out using GraphPad Prism (GraphPad 6 Software, San Diego, CA). Probabilities lower than 0.05 were considered statistically significant (*p < 0.05; ***p < 0.001, ****p < 0.0001).

RESULTS

The present study aimed at characterizing histopathological alterations with a special focus on neural injury and repair in a novel intranasal LM infection mouse model mimicking LOD. In H&E-stained brain sections of all mice exposed to LM, meningitis or meningoencephalitis was diagnosed by detection of infiltrating white blood cells in the meninges and/or the brain parenchyma (Fig. 1A, B). Ischemic injuries and axonal damage were not found in the brains of infected or uninfected mice.

Two approaches were chosen to detect apoptotic cells in the dentate gyrus of the hippocampal formation: in situ tailing in combination with morphological criteria (condensed, shrunken nuclei, condensed and shrunken eosinophilic cytoplasm) for the detection of fully developed apoptosis and in situ tailing in combination with staining for activated caspase 3 visualizing also early stages of apoptosis. In situ tailing plus morphology revealed that infection of the brain with LM induces apoptotic neural injury (LM-infected mice: median = 77.7/mm²; control mice: median = 55.0/mm²) (p = 0.044) (Fig. 2A). In line with this finding, a significantly higher density of neurons labeled by activated caspase 3 in combination with in situ tailing was found in brains of LM-infected mice (median = 121.4/mm²) compared to control mice (median = 74.4/mm²) (p < 0.0001) (Fig. 2B).

FIGURE 2.

Densities of apoptotic and immunoreactive cells and activity of microglia in mice infected with LM and control mice. In sections of the hippocampal dentate gyrus, more apoptotic neurons (A, in situ tailing plus morphology; B, in situ tailing plus caspase 3-staining) were found in mice with meningitis or meningoencephalitis compared to control mice. Lower densities of dividing cells (proliferating cell nuclear antigen [PCNA] staining) (C) and young neurons (D, calretinin and E, calbindin staining) were detected in LM-infected mice. The density of microglia was increased in LM-infected mice compared to controls (ionized calcium binding adaptor molecule 1 [Iba-1] staining) (F). Morphological scoring of microglia revealed no differences in both groups, which is probably a consequence of basal microglial activation in neonatal control mice (G). The density of astrocytes was similar in both groups (glial fibrillary acidic protein [GFAP] staining) (H). (A, C, D) n = 14 LM-infected mice and n = 14 control mice; (B) and (F–H) n = 14 LM-infected mice and n = 13 control mice; (E) n = 13 LM-infected mice and n = 12 control mice. *p ≤ 0.05; ****p ≤ 0.0001.

The density of proliferating cells stained by PCNA as a marker of regeneration was lower in the subgranular zone and granule cell layer of the dentate gyrus in LM-infected mice (median = 369.7/mm²; control mice: median = 556.1/mm²) (p < 0.0001) (Fig. 2C). In support of these findings, infection of the brain with LM reduced the density of calretinin-positive cells (LM-infected mice: median = 5034/mm²; control mice: median = 6525/mm²) (p < 0.0001) and calbindin-28-D-positive cells (LM-infected mice: median = 1328/mm²; control mice: median = 1495/mm²) (p = 0.011), two markers which label young neurons (Fig. 2D, E).

Microglial cells and astrocytes, two cell types that are known to play key roles against infections of the CNS (21, 22), were analyzed in the subgranular zone, granular cell layer, and the hilus of the dentate gyrus. The density but not the activation score of microglial cells was found to be significantly higher in the brains of LM-infected mice (density: median = 92.8/mm²; activation score: median = 1.75) compared to controls (density: median = 53.0/mm²; activation score: median = 2) (density: p < 0.0001; activation score p = 0.49) (Fig. 2F, G). The density of astrocytes was similar in both groups (LM-infected mice: median = 55.1/mm²; control mice: median = 50.8/mm²) (p = 0.78) (Fig. 2H). LM was detected by immunofluorescence microscopy in proximity to infiltrating cells (Fig. 3).

FIGURE 3.

Detection of LM in brain sections of neonatal mice. Nuclei were stained with DAPI (blue; A and C) and LM with a polyclonal anti-LM antibody (red; B and C). The area shown in Figure 1B was stained for the detection of LM.

DISCUSSION

CNS infections in newborns often affect cerebral maturation and neurodevelopment (12). In children surviving bacterial meningitis, disabilities can persist life-long (13–16). The nasal LM infection model described by Pägelow et al 2018 appears to be ideal to test potential therapies that could improve the outcome of LOD in neonates since this model induces infection of the CNS in all mice exposed to the pathogen. It also utilizes the suspected natural route of infection (10). However, a detailed characterization of the histopathological alterations in this neonatal brain LM model was missing.

In the present study, we focused on the detection of neural injury and repair in the hippocampal formation, a brain region that is vulnerable to a variety of noxious stimuli including inflammation and which is essential for memory and spatial learning (23). In rodent models, learning impairments caused by bacterial meningitis are connected to apoptosis of dentate granule cells in the hippocampal formation (17, 18). In human autopsy cases of bacterial meningitis and experimental animal meningitis models using group B streptococcus, Streptococcus pneumoniae and Escherichia coli, neuronal apoptosis in the hippocampal dentate gyrus is a prominent finding as well (24–27). Taking the aforementioned literature into account, we decided to focus on neuronal injury and repair in the present study. Two staining approaches were chosen to detect apoptotic cells in the dentate gyrus of the hippocampal formation: in situ tailing combined with morphology (shrunken and condensed cytoplasm) to visualize the full extent of cell apoptosis, and in situ tailing plus staining for activated caspase 3, an enzyme that is present in apoptotic cells, coordinates degradation of cellular structures (28) and also labels early stages of apoptosis in which the cytoplasm is still detectable (29). Both methods revealed a higher density of apoptotic neurons (identified by location and morphology) in LM-infected mice compared to control animals. The density of positive cells was higher after labeling cells with caspase 3 plus in situ tailing (LM-infected mice: median = 121.4/mm²; control mice: median = 74.4/mm²) compared to in situ tailing plus apoptotic morphology (LM-infected mice: median = 77.7/mm², control mice: median = 55.0/mm²), since caspase 3 in addition labels early stages of apoptotic cells. The density of apoptotic neurons in the dentate gyrus was comparable to a neonatal rat group B Streptococcus model using intracerebral infection (24), and adult rabbit models of S. pneumoniae and E. coli meningitis using intracisternal infection (25, 26). However, it was higher compared to an adult mouse model of pneumococcal meningitis using intracerebral inoculation (30, 31), and an intranasal Streptococcus suis infection model in growing piglets (meningitis group: 0–30 apoptotic cells/mm² vs control group: 0–3 apoptotic cells/mm²) (32).

Neural injury observed in this model might be induced by the secretion of bacterial factors, such as the cytotoxin listeriolysin O (LLO). Although LLO exerts the highest activity at an acidic pH of 5.5 which is present in phagosomes, cytotoxic activity is also detectable at a higher pH and was found to cause neuronal apoptosis (33). In line with our results, Schlüter et al (34) demonstrated that intracerebral infection of mice with LM is accompanied by neuronal apoptosis in the hippocampus and the cerebellum; they assumed that LLO exerts this effect. Furthermore, activation of microglial cells by Gram-positive cell wall components or other bacterial products can induce neuronal damage (35–39). In the present study, we were unable to detect a higher activation of microglial cells in the brains of infected mice compared to uninfected controls by morphological criteria. In order to identify the 5 steps of microglial activation described in Materials and Methods as precisely as possible, microglial activation was quantified by morphological criteria adapted from Kreutzberg et al. (20). To reduce inter-person and inter-day variability, scoring was performed by a single blinded observer (M.B.) within 1 day. Manual or automated analysis of microglial morphology based on immunohistochemically stained sections provides valuable information about microglial activation. For the study of microglial function, however, additional methods, such as flow cytometry, transcriptomics, and immunohistochemistry for cyto- and chemokines are necessary (40). In both groups, we found signs of microglial activation such as shortened processes with an increased thickness. In the mature brain, microglial cells are ramified but in the developing brain, these cells are activated (41, 42) to fulfill tasks such as synaptic pruning (43). For this reason, the present model appears unsuitable to detect microglial activation by morphological criteria. Conversely, the density of microglial cells was significantly higher in LM-infected mice compared to controls, suggesting microglial proliferation in response to the inflammatory stimulus. In the original description of this model, Pägelow et al were unable to detect an increased number of microglial cells in the infected mice by flow cytometry. This deviation from our results probably is a consequence of methodical differences: Pägelow et al detected microglia cells using other markers (CD45^loCD11b⁺CX3CR1⁺Ly6C^-) and counted microglial cells in the whole brain, whereas we measured the density of Iba1⁺ cells in the dentate gyrus of the hippocampal formation, the most vulnerable region of the brain in meningitis. In line with our results, a higher density of microglial cells was observed in human autopsy cases, in which pathogen-induced inflammation caused microglial proliferation (19, 44).

In animal models of bacterial meningitis using growing piglets, adult mice, and rabbits an increased neurogenesis is observed (31, 32). Similarly, in adult human autopsy cases of bacterial meningitis, septic metastatic encephalitis and fungal meningoencephalitis neural repair is often detectable in the hippocampal formation (44–46). In contrast, the densities of PCNA-, calretinin, and calbindin-positive cells were significantly decreased in LM-infected mice compared to control animals indicating that LM infection of the neonatal CNS reduces neurogenesis. In line with our results, a reduced neurogenesis was found in the hippocampus of neonatal piglets suffering from porcine reproductive and respiratory syndrome virus infections (17). In neonatal rats infected with the lymphocytic chorion meningitis virus (LCMV), a loss of dentate granule cells was detected which was accompanied by a loss of neuronal progenitor cells thereby leading to short- and long-term impairments in brain development (47). In addition, congenitally acquired LCMV infection decreases neuronal progenitor cells in the hippocampus and the subventricular zone in adult mice (48). This discrepancy probably is explained by the age of the animals used in the different models: in adult animals and humans, hippocampal injury in bacterial meningitis induces repair mechanisms of moderate intensity, whereas in newborn animals, bacterial meningitis injures neural progenitor cells thus impairing regenerative activity. This hypothesis is supported by findings of Hofer et al (49), who demonstrated that bacterial meningitis injures hippocampal stem and progenitor cells in an infant rat model and thus impairs hippocampal neurogenesis. It remains to be determined whether this observation in newborn animal models also is valid for human neonates and infants. Impairment of neural progenitor cells in neonatal bacterial meningitis could explain the severe sequelae and developmental deficits in children and adults surviving neonatal bacterial meningitis.

Astrocytes are the most abundant cell type in the CNS and are thought to play key roles in infections of the CNS (22). An increase in the density of astrocytes was observed in the hippocampal formation of human autopsy cases of HIV encephalopathy (19), but not in autopsies of bacterial meningitis (45). In adult mice after inoculation of bacteria into the brain, reactive astrocytes were observed close to the infection site (30). In adult transgenic mice after injection of S. pneumoniae in the subarachnoid space, an increase in GFAP activity detected by bioluminescence as an indicator of activation of astrocytes was found. Immunohistochemically, GFAP activity in brain sections from infected animals was multifocally increased in regions bordering necrotic areas (50).

In the present study, we did not find differences in the density of astrocytes between infected and uninfected mice. This could be a consequence of the short period of time between infection and organ sampling. Similarly, axonal damage and ischemic injury were also not observed, although microbial compounds can cause axonal damage by stimulation of immune cells and by vasculitis (36, 46, 51), and ischemic injuries are detectable in meningitis cases in adult humans (27, 52). To detect axonal injury and ischemic lesions, this model will have to include antibiotic treatment or a reduced number of inoculated bacteria, which will allow study of mice at larger intervals between infection and organ sampling.

CNS infections in newborns often affect cerebral maturation and neurodevelopment. Until recently, there was a lack of animal models that use the natural route of infection and thereby reproduce the pathogenesis of late-onset neurolisteriosis in human neonates. The model developed by Pägelow et al in 2018 (10) uses nasal inoculation of bacteria and induces infection of the CNS by colonization of the nasal mucosa, invasion of the olfactory neurons and the olfactory bulb, spreading from there to other brain regions without using the hematogenous route of infection. The morphological alterations observed in the brains of neonatal mice with LM meningitis or meningoencephalitis, regarding the increase in microglial density and neural injury in the hippocampal formation, resemble those observed during bacterial meningitis in adult humans and infants as well as in animal models. A key finding of this study is that cell proliferation and neurogenesis is decreased during neurolisteriosis in neonatal mice, in contrast to an increased neurogenesis in adult animal models or adult human autopsy cases. With the knowledge of morphological alterations caused by neurolisteriosis in this model, novel therapies aiming at reducing brain injury and learning deficits in neonatal bacterial meningitis can be tested in the future.

The authors report no competing interest regarding the contents of the article.

ACKNOWLEDGEMENTS

We thank Cynthia Bunker for careful language editing.

This study was supported by the B. Braun-Stiftung. The funding bodies did not influence the design of the study, collection, analysis, and interpretation of data and the writing of the manuscript.

Supplementary Data can be found at academic.oup.com/jnen.

Author notes

Jana Seele and Melissa Ballüer contributed equally to this work.