Spatio-temporal brain invasion pattern of Streptococcus pneumoniae and dynamic changes in the cellular environment in bacteremia-derived meningitis

Highlights  3D whole brain imaging for the visualization of brain invasion by blood-borne bacteria.  Invasion pattern of pneumococci in brain regions and in neurons’ plasma membrane.  Dynamic response of the dentate gyrus neurogenic niche during brain infection. Abstract Streptococcus pneumoniae (the pneumococcus) is the major cause of bacterial meningitis globally, and pneumococcal meningitis is associated with increased risk of long-term neurological sequelae. These include several sensorimotor functions that are controlled by specific brain regions which, during bacterial meningitis, are damaged by a neuroinflammatory response and the deleterious action of bacterial toxins in the brain. However, little is known about the invasion pattern of the pneumococcus into the brain. Using a bacteremia-derived meningitis mouse model, we combined 3D whole brain imaging with brain microdissection to show that all brain regions were equally affected during disease progression, with the presence of pneumococci closely associated to the microvasculature. In the hippocampus, the invasion provoked microglial activation, while the neurogenic niche showed increased proliferation and migration of neuroblasts. Our results indicate that, even before the outbreak of symptoms, the bacterial load throughout the brain is high and causes neuroinflammation and cell death, a pathological scenario which ultimately leads to a failing regeneration of new neurons.


Introduction
Pneumococcal meningitis, the inflammation of the meninges caused by the Gram-positive bacterium Streptococcus pneumoniae (the pneumococcus), is a medical emergency, with a mortality ranging between 18 and 30%, depending on the geographical region, with half of survivors suffering from long-term neurological sequelae, most commonly hearing loss, motor disabilities and cognitive impairments (Engelen-Lee et al., 2016).In the majority of pneumococcal meningitis cases, invasion into the brain and meninges follows a hematogenous route of infection, although translocation from the nasopharynx through the cribriform plate or through direct implementation in the brain following trauma or surgery does occur (Gil et al., 2023;Lin et al., 2014;Audshasai et al., 2022).The presence of S. pneumoniae in the brain causes a strong inflammatory response, with release of reactive oxidative species and matrix metalloproteinases that augments pathological progression (Klein et al., 2006).A consequence of the neuroinflammation is the increased intracranial pressure due to edema, which is linked with increased risk of cerebral infarction and ischemic injury in patients (Lindvall et al., 2004).This, together with the direct toxicity of pneumococcal virulence factors, cause neuronal damage and death (Gil et al., 2023;Schut et al., 2012).
Experimental models of pneumococcal meningitis have shown that S. pneumoniae penetrates the blood-brain barrier (BBB) in a transcellular and paracellular pathway (Gil et al., 2023).Several receptors have been described to be utilized by the pneumococcus during transcytosis through endothelial BBB cells.This includes the laminin receptor which binds pneumococcal expressed PspC (Orihuela et al., 2009).PspC also binds platelet endothelial adhesion molecule (PECAM)-1 and the polymeric immunoglobulin receptor (pIgR) on brain vascular endothelial cells (Cundell et al., 1995;Iovino et al., 2017).Furthermore, pneumococci express enolase, which binds plasminogen that is expressed on endothelial cells, facilitating bacterial adhesion (Bergmann et al., 2013).Pneumococcal phosphorylcholine has been shown to adhere to the platelet endothelial adhesion factor (PAF) receptor facilitating transcytosis (Cundell et al., 1995).On the other hand, the survival of S. pneumoniae during the transcellular passage has been shown to be partly dependent on the pneumolysin (Ply) expression (Surve et al., 2018).While the majority of S. pneumoniae are eliminated during this transcellular passage, some survive and are either transported towards the basal side of the endothelium or back to the apical side (Ring et al., 1998).In addition, the inflammatory response of the endothelial cells in response to pneumococcal virulence factors, such as neuraminidase A, increases the adhesion and translocation of the pneumococcus across these cells (Banerjee et al., 2010).A paracellular pathway involving PECAM-1 has been observed, and BBB breakdown is prevalent in patients, which further facilitate a paracellular route of entry (Iovino et al., 2016).The pneumococcus has been shown to initially invade through the BBB of vessels close to the subarachnoid space in early stages of disease and, in later stages, to reach the inner brain through the choroid plexus where it can then likely enter the cerebrospinal fluid (Iovino et al., 2013).This is in contrast to Streptococcus suis, where the choroid plexus has been reported as the main point of entry, while histopathological examination from Neisseria meningitides meningitis show equal bacterial entry through both the microvasculature and the choroid plexus (Schwerk et al., 2015).Even though the molecular interactions between pneumococci and brain endothelial cells have been characterized, it is still poorly understood whether pneumococcal entry of the brain occurs preferentially in certain brain regions (Gil et al., 2023;Iovino et al., 2016).
Once in the brain, S. pneumoniae interacts with neurons by binding to the β-actin exposed on the neuronal plasma membrane through its pilus-1 tip protein RrgA, and through this interaction pneumococci are thought to invade neurons, ultimately causing neuronal death through cytoskeleton disruption (Tabusi et al., 2021).Pneumococci produce also H 2 O 2 ; together with the pore forming toxin Ply, it contributes to the cellular death of neurons and astrocytes, while at the same time inhibiting the motility of microglia, thus reducing phagocytosis (Farmen et al., 2021).Microglia sense S. pneumoniae through extracellular receptors such as Toll-like receptors 2,3,4, and 9, as well as through intracellular receptors including nucleotide-binding oligomerization domain like receptors (Ribes et al., 2010;Paterson and Mitchell, 2006).This recognition causes the release of extracellular cytokines and chemokines, which drives the subsequent infiltration of neutrophils into the brain (Yau et al., 2018).While innate immune cells are capable of phagocytosing the bacterium, this clearance is ineffective, causing an upheld proinflammatory state, release of oxidative species, and increased neuronal damage (Mook-Kanamori et al., 2011).Interestingly, neuroinflammation and neuronal damage can also occur without bacterial invasion into the brain.In fact, Orihuela et al., showed that, during bacteremia with no pneumococci present in the brain parenchyma, microglia were activated and neuronal damage in the hippocampal dentate gyrus was detected (Orihuela et al., 2006).In patients that died of bacterial meningitis, apoptotic neurons in the dentate gyrus of the hippocampus were observed in the majority of cases (Gianinazzi et al., 2003).The subgranular zone (SGZ) of the dentate gyrus represents one of the two main neurogenic niches in the adult brain.Newly generated cells in the SGZ can become integrated in the neuronal circuits involved in cognition, memory, and learning.Damage to these pathways has been shown to cause behavioral impairment (Abbott and Nigussie, 2020;Baker et al., 2016).This has also been observed in experimental pneumococcal meningitis, where caspase-3 dependent apoptosis in the hippocampus was linked with learning deficiencies in the rodents (Gianinazzi et al., 2003).Thus, the observed neurological sequelae in patients are a consequence of several pathophysiological processes and dependent on the severity and the brain regions affected.
While treatment with anti-inflammatory drugs in rodents attenuates brain damage, and dexamethasone treatment in patients has shown to be beneficial in terms of outcome (van de Beek et al., 2004;Barichello et al., 2016), there is still no treatment available to patients that attenuates the observed neurological sequelae significantly.In experimental models, BDNF and anti-oxidative treatments have both shown promise, as well as inhibitors of MMPs; however, these remain to be available for clinical testing (Mook-Kanamori et al., 2011).Partly, the difficulties in finding therapeutics that reduce neurological sequelae in patients are due to the lack of understanding of the disease progression and the variability of the experimental animal models used.Mainly two in vivo experimental mouse models are used in the field, the haematogenous meningitis model and the intracisternal meningitis model (Iovino et al., 2017;Schlüter et al., 1996;Koedel et al., 2001;Leib, 2001;Tan et al., 1995;Orihuela et al., 2004;Radin et al., 2005).The main strength of the bacteraemia derived model is that it follows the natural route of infection in human patients, and thus it is relevant to be used when investigating pneumococcal translocation across the BBB and invasion of the brain.However, as a high degree of bacteraemia also is developed during the course of the disease, the clinical symptoms of the mice and the pathophysiology will be affected.In contrast, the intracisternal model bypasses the crossing of the BBB and induces a compartmentalized meningitis.Disease pathophysiology in this case, then, is provoked by meningitis exclusively; retrograde crossing of the bacteria from the brain to the periphery is prevented by the use of antibiotics.In addition to the route of administration, the choice of mouse strain, bacterial strain and dosage will also impact the experimental outcome (Chiavolini et al., 2008;Borsa et al., 2019).
In this work, our goal was to utilize the bacteraemia-derived meningitis mouse model to describe the tempo-spatial invasion pattern of pneumococci into the brain.Furthermore, we aimed to investigate the impact of this invasion on neuronal apoptosis, microglial activation, and neurogenesis.Our results show that the pneumococci invade all brain regions in an equal manner; to our knowledge, we are the first to visualize bacterial invasion in the brain utilizing light sheet microscopy of three dimensional (3D) whole brain imaging.This bacterial invasion caused activation of microglia and with the neurogenic niche showing an initial increase in neuroblast proliferation and migration.This study shows the dynamic changes in the brain cellular environment during pneumococcal meningitis pathogenesis over time, following the progression of bacterial invasion into the brain.

Methods
The key resources described below are summarized in the Key Resources Table.

Animal experiments
Five-to seven-week-old male C57BL/6 mice were supplied from Charles River and housed in a 12 h light/dark period with ad libitum access to water and food, in accordance with Swedish legislations (Svenska jordbrukseverket, ethical permit number 18965-2021 and13,890-2022).We utilized a bacteraemia derived meningitis model as previously described (Orihuela et al., 2009).Shortly, each mouse received an intravenous injection of 1 × 10 8 colony forming units (CFU) of S. pneumoniae in PBS solution through the tail vein; control animals received a PBS injection.Mice were checked for clinical symptoms every 3 h and scored according to Karolinska Institutet "Assessment of health conditions of small rodents and rabbits when illness is suspected" ("Bedömning av djurhälsa för smågnagare och kanin vid misstänkt ohälsa") template.Shortly, mice were scored 0.1 if they had one of the following symptoms: reduced activity, slow reflex, piloerection, or squinty eyes.A score of 0.2 was given if the mice showed reduced activity and mobility.At each check-up, the scores were summed together.Mice were divided into three groups: 2 h post infection (asymptomatic), mild symptomatic (score of 0.1/0.2) and severe symptomatic (score 0.3/ 0.4).Supplementary Table 1 summarizes the average and range of time of sacrifice after injection of S. pneumoniae for each experimental group.Additionally, for each time point, mice were separated into three groups for downstream analysis: 1) Brain microdissection and CFU count (n = 5), 2) Serial sectioning for immunofluorescence analysis (n = 3) and 3) iDISCO protocol for whole brain imaging (n = 1).At time of sacrifice, 5 μl of blood was collected from the tail vein before the mouse was perfused with ice cold PBS through the left ventricle and the spleen collected for analysis of systemic infection; for analysis of systemic infection n = 12-17.The brain was carefully collected and placed in icecold 4% PFA for downstream analysis 2 and 3.For group 1, the brain was micro-dissected into seven brain regions (striatum, hippocampus, frontal cortex, cortex, midbrain, and cerebellum).The regions were placed in ice cold PBS, weighed, and then homogenized through a 30 μm cell strainer.Bacterial presence in blood, spleen, and brain was quantified by serial diluting, plating the bacteria onto blood agar plates, and CFU count.The same model was used in a separate in vivo experiment to assess brain invasion by pneumococci in the presence of antibiotics.Ceftriaxone (Navamedic) 100 mg/kg was given I⋅V 8 h after injection of S. pneumoniae, while vehicle animals (infected) received PBS I.V.When an infected vehicle mouse reached severe classification (0.4 score) one mouse from ceftriaxone group was also sacrificed regardless of the score of the antibiotic treated mouse.8 h was chosen as this was just prior to animals in the first experimental group starting to develop a score of 0.1.Sampling, tissue harvesting, and microdissection was performed as previously described and were analyzed by CFU count.

Preparation of bacterial strain
Streptococcus pneumoniae TIGR4 (serotype 4) laboratory strain and a clinical isolate serotyped 6 A (Iovino et al., 2017) were grown to an OD of 0.3-0.4 in Todd-Hewitt broth (THY) and stored in a solution containing 20% glycerol and THY at − 80 • C.

Embryonic mouse primary neuron isolation, culture, infection, and immunofluorescence
Mouse primary cortical neurons were prepared from the cortex of E18-old embryos isolated from C57BL/6 pregnant mice as previously described (Viesselmann et al., 2011), and in accordance to Swedish legislations (Svenska jordbruksverket, ethical permit number 17038-2020).Briefly, brains were carefully isolated from embryos, dissected and the cortical tissue triturated with fire-polished Pasteur pipettes.The isolated cell suspension was seeded on coverslips in 12well plates (1 × 10 5 cells/well) coated with L-ornithine.Cells were cultivated in Neurobasal™ medium (Thermo Fisher Scientific, USA) supplemented with 1% B-27 Plus Supplement (Thermo Fisher Scientific, USA) and penicillin-streptomycin 1%.Half of the media was changed every three days.Cells were cultured for 15 days.The day before the experiment, media was changed to the same without antibiotics.Then, cells were either infected with S. pneumoniae TIGR4 at multiplicity of infection (MOI) 10 for 2 h or treated with Ply (Protein Production Platform, Singapore) 0,02 μg/mL for 30 min.Infected cells were washed twice with PBS at 37 • C to eliminate unbound bacteria, and fixed with 4% PFA in PBS for 15 min at room temperature, blocked with 1% BSA in PBS for 20 min and incubated with different primary antibodies.For bacteria localization, a rabbit antiserum against serotype 4 S. pneumoniae capsule (SSI Diagnostica, Denmark) at 1:100 was used, followed by a goat-anti rabbit IgG Alexa Fluor 594 secondary antibody at 1:500.Slides were mounted in Prolong Gold antifade reagent (Molecular Probes, Invitrogen) and examined with a confocal microscope (Zeiss LSM900-Airy) in combination with differential interference contrast (DIC) microscopy at 63× magnification.For β-actin analysis, monoclonal IgG antibodies against β-actin (Invitrogen) and L1CAM (R&D Systems, USA) at 1:100, mouse and rabbit respectively, were used, followed by secondaries goat anti-rabbit IgG Alexa Fluor 594 and goatanti mouse IgG Alexa Fluor 488 at 1:500.Slides were mounted in Prolong Gold antifade reagent (Molecular Probes, Invitrogen) and examined with a confocal microscope (Stellaris 5, Leica) at 63× magnification.

Immunostaining sections
Brains were post-fixed in 4% PFA for 16 h and placed in 30% sucrose solution for minimum 4 days, before coronal sections of 30 μM were cut using a microtome, and the sections frozen in a cryoprotective solution (30% sucrose, DMSO and PBS).For each analysis, three sections per mouse brain were used.The sections were washed in PBS, blocked with 2.5% BSA and permeabilized with 0.3% Triton-X-100 (Sigma) for 1 h at room temperature and stained with primary antibodies overnight at 4 • C.After washing with PBS, the sections were incubated with secondary antibodies for 2 h room temperature; nuclear staining was done by adding DAPI (Abcam) for 10 min.The primary antibodies used in this study were: Iba1 (Abcam, 1:1000), pneumococcal protein CCrZ (In house, 1:100), B 3 -tubulin (Promega,1:1000), Ki-67 (Abcam, 1:200), NeuN (Merck Millipore,1:250) and caspase-3 (BDpharmingen, 1:250).Secondary antibodies used were Alexa Fluor conjugated antiimmunoglobulin at 1:1000 (goat anti-rabbit IgG Alexa Fluor 594 and goat-anti mouse IgG Alexa Fluor 488).The microvasculature was stained using Lectin-488 conjugate.Images was obtained taking a z-stack containing between 17 and 20 sections of 1 μM thick.Images were taken at both 20× and 63× magnification on Zeiss LSM900-Airy, LSM800-Airy and LSM880 confocal microscope, with settings kept constant between respective image acquisition used in quantification.Images acquired for assessing caspase-3 intensity were taken on an Axio serial scanner with a magnification of 20×.

Tissue clearing and light sheet microscopy
For tissue clearing, the iDISCO protocol was used (https://idisco.info/idisco-protocol/) (Renier et al., 2014).The primary antibody used was pneumococcal serotype 4 anti-capsule (SSI Diagnostica, 1:250), secondary antibody used was Alexa Fluor 620 (1:1000).Whole brain images were acquired by a LaVision Biotec (UltraMicroscope II) lightsheet microscope and Imspector software.The whole brain was imaged in a total of 5 mm depth, with the laser lines 488 (autofluorescence) and 639 (bacteria staining) used.The images of the autofluorescence of the brain Were recorded with a resolution of x = 4.797 and y = 4.796, and a z-step size of 4 μm, captured with 1.26× magnification.For bacterial staining of the brain, the x and y resolution were 0.755 and a z-step size of 4 μm, captured with 8× magnification.
The xy images were tiled with a 10% overlap and stitched using Imspector and TeraSticher to generate 2D images.

Image processing and analysis
Amira (ThermoFisher™) was used to analyse the 3D whole brain imaging data.The autofluorescence images were sharpened and the noise reduced by using the following plug-ins: gaussian filter, unsharp masking and brightness contrast.Then, the data was segmented by applying a threshold that was determined manually.Volume rendering of the signal generated a signal of the brain, used as a "skeleton" for the subsequent overlay of bacterial signal.Images acquired from the bacterial staining were downsized to match the autofluorescence images, before the images where sharpened and background reduced by using unsharp mask and background detection correction plug in.Segmentation of the bacterial signal was done by placing a manual threshold to the images, before segmentation toolbox was used to remove noise and artifacts.In all images, the outer edge of the brain was removed as this area contained a mix of high signal due to antibody unspecificities and bacterial staining (Supplementary Fig. 1).The two label fields were overlaid and animated in Amira the videos were generated using DaVinci Resolve.
ImageJ (Fiji) was used to analyse the confocal and slide scanner images.To assess the intensity of Iba1 staining, the 20× magnification stacks were merged, background subtracted, and image converted to binary image by threshold, outliers were removed and with finally outline plug in was used to assess the staining of cells only.The intensity was measured using the measure plugin.To assess numbers of pneumococci in the cortex, we manually counted the number of CCrZ positive cells that co-localized with, or not, the microvasculature.The caspase-3 and NeuN double positive cells was counted in whole coronal sections.To assess preference of bacterial interaction with primary neurons, relative area of cell body vs. projections was measured in each field.
Counting of cells was done blinded.

Statistical analysis
GraphPad Prism-9 was used for statistical analyses (α5%).Outliers were identified by Grubb's test.Normality was tested with D'Agostino-Pearson test.Two-way ANOVA was used to assess differences in CFU in brain areas at different groups.While Ordinary one-way ANOVA or Kruskal-Wallis test was performed with Tukey or Dunn's multiple comparisons as indicated in figure legends.

S. pneumoniae invades the brain, equally affecting all brain regions and causing neuronal damage
The majority of pneumococcal meningitis cases follow bacterial invasion into the brain from the systemic circulation; thus, we utilized the established bacteremia-derived meningitis model to mimic the etiology of the disease (Orihuela et al., 2009;Iovino et al., 2017;Tabusi et al., 2021).To describe the progression of the disease, we divided mice into four groups control (PBS injected), asymptomatic, mild, and severe symptomatic.The piliated S. pneumoniae strain TIGR4 was chosen to generate pneumococcal meningitis as it is highly virulent in mouse models (Mitchell and Mitchell, 2010).To evaluate the spatial invasion pattern into the brain, 3D whole brain imaging and CFU count of seven brain regions were performed.After inoculating the mice with TIGR4, the disease progressed rapidly with increased numbers of bacteria in the blood (Fig. 1A) and spleen (Supplementary Fig. 2 A) observed from the asymptomatic to severe symptomatic group.No bacteria were observed in any tissue sample taken from controls (data not shown).The number of pneumococci also increased in the brain with time, with each of the seven brain regions equally affected by the invading pneumococci as visualized by 3D whole brain imaging (Fig. 1C, videos in Supplementary) and by CFU count of dissected brain regions (Fig. 1B).Importantly, we confirmed that this invasion pattern was also observed for the piliated clinical isolate 6 A, although lower bacterial counts in the brain and reduced virulence of the strain were observed (Supplementary Figs. 3  A and 3B), a feature not uncommon for strains isolated from human meningitis patients that has never previous passaged in vivo.Our results therefore indicate that piliated pneumococci invade the brain without a spatial difference, which is in unison with the bacteria employing the BBB as a primary site of invasion.Because these mice experienced a high degree of bacteremia in addition to the brain infection, we also addressed if the bacterial invasion pattern was altered when antibiotics were given during disease progression.Prior to the development of mild symptoms corresponding to eight hours after infection, ceftriaxone was administered systemically.Vehicle mice received PBS injection and were sacrificed when they developed severe symptoms and ceftriaxone treated mice were also sacrificed, regardless of their clinical symptoms.The ceftriaxone treatment caused a significant reduction in the number of S. pneumoniae in the blood and spleen (Fig. 1D and Supplementary Fig. 2B).In the brain, the number of bacteria in the different brain regions was not different (Fig. 1E), although we observed a significant reduction in the numbers compared to the vehicle mice.
Our visualization of the pneumococci using 3D whole brain imaging showed that a significant amount of bacterial signal was detected around and in the microvasculature of the brain.We therefore sought to investigate further pneumococcal invasion into the brain parenchyma.In the cortex, we observed significantly higher numbers of pneumococci with time, when comparing asymptomatic, mild, and severe mice (Fig. 2A).When quantifying the co-localization between the microvasculature and pneumococci, we observed a significant increase in bacteria located in the parenchyma compared to the vasculature as disease progressed (Figs.2B and C).The number of vasculature-associated pneumococci also increased with time (Fig. 2D).No significant difference was observed between the ratio of parenchymal/total nor vasculature/total bacteria between the three groups (data not shown), indicating that the translocation from the BBB into the parenchyma was constant and dependent on the number of total bacteria in the blood.However, in all animal groups, most bacteria were observed in the vasculature compared to the parenchyma.Bacterial infiltration into the brain caused increased caspase-3 signal in neurons, most prominent in the severe symptomatic mice, indicating caspase-3 dependent apoptosis (Fig. 2E).Thus, our results indicate that the pneumococci invade into the parenchyma during all disease stages, and this is associated with a time dependent increase in apoptosis among neurons.

S. pneumoniae preferably adheres to the cell body of primary cortex neurons
After analyzing the S. pneumoniae pattern of brain invasion in the different brain regions during disease's progression, we wanted to investigate whether pneumococci have a preferential adhesion pattern on neurons at a cellular level.We have previously shown that piliated S. pneumoniae adheres to β-actin exposed on the cell surface of a differentiated neuron-like neuroblastoma cell line (Tabusi et al., 2021).To further elaborate on this process, we investigated if TIGR4 pneumococci showed a preferential spatial binding site on primary cortex neurons.Our results indicate that TIGR4 shows a preference towards the cell body of the neuron (Fig. 3A).As neuronal actin structures are notoriously different in different parts of the cell, we then assessed the presence of exposed β-actin on the plasma membrane.We saw that exposure of β-actin on the neuronal plasma membrane was enhanced after bacterial infection with a significantly increased presence in the cell body of neurons (Fig. 3B).When treating cells with Ply, we observed a strong increase in the exposure of β-actin (Fig. 3C), providing further indications that the pneumococci utilize β-actin, exposed on a plasma membrane damaged due to Ply and other virulence factors, to adhere and invade the neuronal cell, as we have previously speculated (Iovino et al., 2017).

Pneumococcal meningitis caused a dynamic microglial response and altered the neuronal progenitor cell population in the hippocampus
Neurological sequelae in patients with pneumococcal meningitis have been linked to neuronal death in the hippocampal area (Nau et al., 1999).Therefore, we investigated the dynamic changes of microglial activation and neurogenesis in the hippocampus during disease progression.Microglial response in the hippocampus was assessed using the Iba1 marker.The intensity of Iba1 staining decreased in asymptomatic, mild symptomatic and severe mice in comparison to control mice (Fig. 4A).We also observed morphological differences in the four experimental groups.While the controls showed microglia with ramified processes and a small cell body, the severe group showed microglia with more dense protrusions and a more amoebic shape (Fig. 4B).Thus, our results indicate that there is a decrease in Iba1 staining and morphological changes in the microglia as disease severity increase.
Alteration in microglia activation has been shown to influence the neurogenesis in the dentate gyrus of the hippocampus (Ekdahl et al., 2003).We used the neuroblast marker B 3 -tubulin and proliferation marker Ki-67 to evaluate how neurogenesis was affected during disease progression.We observed a decrease in B 3 -tubulin staining in the SGZ in the dentate gyrus, although it did not reach statistical difference.In contrast, this staining was increased in the hilus (Figs.5A-C), suggesting that neuroblasts are migrating towards neuronal circuits.Ki-67 is upregulated in newly proliferated cells and thus, combined with B 3tubulin positive marker, shows evidence of neuroblast proliferation.When quantifying the number of double-positive cells, we observed an initial increase in numbers from control animals to asymptomatic and mild symptomatic mice; however, in severe symptomatic mice, the numbers of double-positive cells decreased and were comparable with control animals (Figs.5C and D).This could indicate that the niche was trying to compensate for the loss of neurons or neuroblast in the area and tried to regenerate, before failing to do so due to the continuous proinflammatory and cytotoxic environment created by pneumococcal products.

Discussion
The World Health Organization (WHO) defines meningitis as "devastating" because, even if the bacterial infection is adequately cured, permanent neurological disabilities occur in approximately half of survivors https://www.who.int/news-room/fact-sheets/detail/meningitis.These long-term neurological sequelae are a burden for patients themselves, but also for health care systems worldwide (Schiess et al., 2021).Importantly, in retrospective studies, children diagnosed and surviving bacterial meningitis show reduced memory and learning abilities in school and throughout life (Chandran et al., 2011;Weisfelt et al., 2006;Glimåker et al., 2015).This highlights the importance of a better understanding of the pathophysiological process during bacterial meningitis.In this study, focused on pneumococcal meningitis, we have described this pathological process in a dynamical way, and we have shown that deleterious events occur early in disease progression with all brain regions affected by pneumococcal invasion.
To our knowledge, we are the first to combine the use of cleared brain tissue and light-sheet microscopy to visualize bacteria in the brain in 3D whole brain imaging; this, together with counting of CFU numbers from brain microdissection, showed that pneumococci invade uniformly into the parenchyma.This heterogenous brain invasion potentially seems to reflect the clinical reality of the neurological sequelae post bacterial meningitis, as the vast variety of sequelae observed are due to damaged motor, sensorial or emotional functions which are controlled by different brain regions (Barichello et al., 2016;Schlüter et al., 1996).Furthermore, in line with previous research from rodent and zebrafish models, the pneumococci were shown to adhere and to be in close proximity with blood vessels, indicating this as the primary site of invasion (Iovino et al., 2013;Jim et al., 2016).We observed a high number of S. pneumoniae in the brain shortly after inoculation of the mice, with bacteria already invading into the parenchyma two hours after infection.While bacterial invasion of the brain progressed, neuronal death also increased accordingly; neuronal apoptosis has been shown to be partially caspase-3-dependent, and an increase in caspase-3 was also observed in our study (Gianinazzi et al., 2003).
Our results support the concept that, once present in the systemic circulation, the bacteria do have the capacity to cross the BBB immediately, and that symptoms of disease are not a consequence of bacterial numbers in the brain, but rather of the subsequent neuroinflammatory response.As expected, and in line with the literature, the pneumococcal invasion caused a morphological change in the microglial cell population, which showed retraction of protrusions and a denser cell body indicating an activated state (Pavan et al., 2021;Zhang et al., 2021a;Iovino et al., 2018).Iba1 is widely used to stain microglia, and we report a decrease in the intensity of this staining; although increased staining intensity is linked with activation of microglia, the exact function of Iba1 is not completely known and it has been reported to be both increased and decreased in intensity in other brain diseases (Lier et al., 2021).Inflammation and microglial activation are detrimental for the Fig. 3. S. pneumoniae preferably adheres to the cell body of primary cortex neurons.A. Pneumococci (capsule antiserum, red channel) adhered with a preference to the cell soma (within white dotted line) rather than the projections of the neurons, imaged with DIC.Each dot represents the average measurement of one biological replicate with, at least, 5 fields of view each, in total 6 biological replicates (63×).Relative area was measured by ImageJ and used to compare results relatively.B. β-actin (green channel) increased its exposure on the plasma membrane and co-localized with L1CAM (red channel) after infection with TIGR4.Each dot represents the average measurement of one biological replicate with, at least, 5 fields of view each, in total 3 biological replicates (63×); neuronal cell soma in infected cells is within white dotted line.C. β-actin (green channel) increased its exposure on the plasma membrane of primary neurons after treatment with Ply 0,02 μg/mL.DAPI (blue channel).Green channel was imaged at high exposure to outline the morphology of the cell by autofluorescence (63×).Data was statistical tested with an unpaired t-test.(For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)generation of new neurons in the dentate gyrus of the hippocampus (Ekdahl et al., 2003;Sheu et al., 2013).In post-mortem brains from bacterial meningitis cases, an increase in neural progenitors was reported (Gerber et al., 2009).These progenitors can differentiate into either astroglia or neuroblasts; and in rodent models of meningitis, a decrease in neuroblasts has been reported (Lian et al., 2016;Segklia et al., 2023).Our results are in line with these previous findings, showing alteration in the proliferation and migration of neuroblasts.We can only hypothesise that the neurogenic niche tries to compensate for a loss of neurons in the early stage of disease, causing an increased proliferation of neuroblasts which is inhibited as the disease progress, but future experiments must confirm this theory.Loss of neuroblasts has previously been shown to withstand for weeks, causing a reduction in neurons, and potentially causing cognitive deficiencies (Hoffmann et al., 2007;Barichello et al., 2014).
The actin cytoskeleton has been linked to bacterial pathogenesis in several ways, from invasion, intracellular motility and navigation to phagocytosis escape (Anes, 2017;Choe and Welch, 2016).Ply has been previously shown to cause dendritic spine collapse via de formation of membrane pores that disrupt the critical cytoskeletal structures that sustain such spines, many of them depending on F-actin processes (Baronti et al., 2023).We have previously suggested a synergy between RrgA and Ply in pneumococcal interaction with neurons; in fact, using purified proteins we have shown that, once in contact with neuron-like cells, Ply alters the plasma membrane increasing the exposure of β-actin, which can therefore be bound by pneumococci, therefore promoting bacterial adhesion (Iovino et al., 2017).Here, we show that the infection with S. pneumoniae causes an increase of β-actin exposure on the plasma membrane of primary murine cortical neurons.This increased exposure was correlated with the interaction of Ply with the plasma membrane.Moreover, confocal microscopy analysis revealed that pneumococcal adhesion preferentially occurred on the cell body of neurons rather than their projections, axon-and dendrite-like structures, most likely because the exposure of β-actin on neuronal plasma membrane was significantly higher in the soma rather than the projections, providing further insights on the mechanism by which pneumococci affect neurons directly.
While several studies have investigated and described the deleterious events in the brain cellular environment during bacterial meningitis pathogenesis, the majority utilizes an intracisternal model (Mook-Kanamori et al., 2012;Paul et al., 2004;Paul et al., 2005).This model has several benefits, including the possibility to use strains without good capability to penetrate the BBB and is, furthermore, the only way to control the exact number of bacteria that enters the central nervous system.However, precisely because it bypasses the initial crossing of the BBB, it does not properly mimic the effect of a systemic infection on the pathological process.In contrast, the bacteraemia-derived meningitis model utilized in this study considers how the progression of the disease occurs in most clinical cases.Furthermore, the effect of systemic infections on the brain environment, and the fact that pneumococci cause damage to the hippocampus before getting access into the brain, underlines the strength of using this model when evaluating the invasion and dynamic cellular changes in the brain environment.However, it should be noted that our results are affected by the presence of bacteraemia in the mice, which has been previously shown to significantly impact the pathophysiology and the clinical symptoms of meningitis (Brandt et al., 2008).In this study, we provide a clear visualization of the invasion pattern of pneumococci into the brain using light-sheet microscopy.Light-sheet microscopy combined with 3D brain imaging has been applied before in literature, however always investigating at specific cells in certain brain regions, such as hippocampal neurons in mouse models (Butt et al., 2021;Nimmo et al., 2020)   communication between large population of neurons in a larval zebrafish model via a more comprehensive whole brain imaging (Ahrens et al., 2013).More recently, studies by Cong et al. and Zhang et al. have adopted whole brain imaging to investigate different neural structures and their respective activities (Cong et al., 2017;Zhang et al., 2021b).
Our study shows for the first time the use of light-sheet microscopy combined with whole brain imaging to monitor invasion of the brain by blood-borne bacterial pathogens during the progression of an invasive infectious disease.Although the bacterium has an average size of 1 μm, we could clearly see bacterial signal in the brain tissue.However, we find it unlikely that these pneumococci represent single bacterium, as this would require higher magnification, currently not available for light-sheet microscopy.Despite this, we believe that this technique will be useful for future investigations of the brain invasion mechanism of invasive pneumococci and other pathogens.
In conclusion, by combining a bacteremia derived meningitis mouse model and light-sheet microscopy for 3D whole brain imaging, we have provided a comprehensive spatio-temporal analysis of brain invasion by blood-borne S. pneumoniae, a crucial process in the pathogenesis of pneumococcal meningitis.In addition, this combination of in vivo model with high-resolution microscopy analysis allowed us to study the inflammatory, neuronal apoptotic and neurogenic response, from the time when pneumococci start entering the brain until the acute phases of disease in which the brain is severely invaded by the pathogen.

Fig. 1 .
Fig. 1.Pneumococcal invasion of the brain.A. Increase in CFU in the blood as disease progresses after IV injection of TIGR4.B. The CFU in striatum, hippocampus, frontal cortex, cortex, midbrain, and cerebellum increased in all brain regions with time.C. Images from 3D whole brain imaging, where the invasion of S. pneumoniae into the brain is visualized.Comparison between ceftriaxone treated mice and non-treated mice (vehicle).A significant decrease in D. Blood and E. Brain CFU in six brain regions of ceftriaxone-treated mice was observed compared to vehicle mice.Bars show median with range.Significance was tested with oneway ANOVA and Tukey's multiple comparison in A) with n = 13-15.D) tested with Mann-Whitney and n = 5.B and E) tested with two-way ANOVA and Tukey's multiple comparison n = 5.

Fig. 2 .
Fig. 2. Pneumococcal invasion into the brain parenchyma causes increased neuronal caspase-3 staining.The presence of pneumococci (CCrZ) in the cortex (lectin) and parenchyma during disease progression in the cortex A. and its quantification in the parenchyma B, and in the vasculature C. Each dot represents the average measurement of region of interest (ROI) in 3 mice (n = 3), in total six ROI.D. Quantification of Caspase-3 bright and NeuN positive cells in coronal sections ratioed to control, n = 6.Bars show median with range.Significance was tested using Kruskal-Wallis test, with Dunn's multiple comparison.E. Representative images for each animal group for the quantification of pneumococci in the parenchyma and the vasculature.Bacteria stained in red, vasculature in green and DAPI in blue.63× magnification.Scalebar indicates 5 μM.(For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.) A step forward in terms of spatial completeness was done by Ahrens et al. that investigated

Fig. 4 .
Fig. 4. Microglial activation in the hippocampus during disease progression.A. Quantification of average mean intensity of Iba1/area.Significance was tested using Kruskal-Wallis test, with Dunn's multiple comparison.Bars show median and range n = 6.B. Representative images from each mouse group; as disease progresses, the microglia show an increase in cell body and decreased ramifications, typical of an activated phenotype.Microglia as shown in red and DAPI in blue.Magnification 63×, Scalebar indicates 15 μM.(For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Fig. 5 .
Fig. 5.The dynamic cellular response of the neurogenic niche in the hippocampus.A. Quantification of β3-tubulin positive cells in the dentate gyrus.B. Quantification of β3-tubulin positive cells in the hilus.C. Quantification of β3-tubulin and Ki-67 double positive cells in the dentate gyrus.D. Representative images from each mouse group, DAPI in blue, Ki-67 in green and β3-tubulin in red.Magnification 63×, Scalebar indicates 20 μM.Significance was tested using Kruskal-Wallis test, with Dunn's multiple comparison.Bars show median with range, n = 6.(For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)