Blood vessel-associated inflammatory microglia and astrocytes are associated with molecular and cellular markers of blood-brain barrier permeability in neonatal mice infected with the respiratory pathogen Bordetella pertussis

Abstract

Severe infection with the respiratory pathogen Bordetella pertussis during infancy has been associated with neurological complications, but the underlying mechanisms are not fully understood. Here we examined the impact of infection of neonatal mice with B. pertussis on the blood-brain barrier (BBB) by assessing pericytes, endothelial cells and tight junctions in cortex. We report that B. pertussis infection was associated with altered spatial distribution of brain pericytes, together with cerebral vasodilation and changes in endothelial cell tight junctions, as well as enhanced vascular deposition of the alarmin S100β. We demonstrate that, while there was an increase in the number of GFAP⁺ blood vessel-associated astrocytes, expression of the end foot marker, aquaporin 4 (AQP4), was decreased in these astrocytes. Microglia from the cortex of infected mice adopted a reactive morphology, migrated towards blood vessels and internalised AQP4 to a greater extent than microglia in un-infected mice. This study indicates that disruption of the BBB, inferred from structural and molecular alterations, during infection of neonatal mice with B. pertussis is linked to activation of vessel-associated microglia and astrocytes. Our findings suggest that the neurological sequelae associated with critical pertussis in infancy may be due, at least in part, to breakdown of the BBB.

Introduction

Bordetella pertussis, the Gram-negative coccobacillus, causes whooping cough (pertussis), a severe respiratory disease that can be fatal in infants and young children. Infants that survive critical pertussis often have impaired cognitive development and learning difficulties post hospital discharge [1]. Pertussis-related encephalitis/encephalopathy, diffuse brain swelling and generalized seizures have been reported in unimmunized children [2]. This has been associated with marked demyelination and elevated levels of myelin basic protein (MBP), interleukin-6 (IL-6) and IL-10 in cerebrospinal fluid (CSF), indicating that immune-mediated mechanisms may contribute to the pathogenesis of neurological sequelae associated with B. pertussis infection in infants.

The blood-brain barrier (BBB) protects the brain parenchyma from infectious insult. However, a number of blood-borne bacteria, including Neisseria meningitidis and Streptococcus pneumoniae, can cross the blood-CNS barriers and cause meningitis [3]. Bacterial pathogens can deploy several mechanisms to promote dissemination and circumvent the BBB. Pertussis toxin (PT) is a key virulence factor secreted by B. pertussis that interacts directly with brain microvascular endothelial cells to increase vascular permeability [4]. Due to its effectiveness at increasing vascular permeability to facilitate paracellular translocation [5], PT is used as an adjuvant to induce experimental autoimmune encephalomyelitis (EAE), a well-established model of multiple sclerosis [6].

Previously, we have shown that respiratory infection of neonatal mice with B. pertussis results in dissemination of bacteria to the brain and this was associated with infiltration of peripheral immune cells, which promoted early postnatal neuroinflammation in the cortex and hippocampus leading to neurodevelopmental deficits and ASD-related phenotypes [7]. The influx of neutrophils, monocytes and activated T cells led us to hypothesize that the structural composition of the BBB may be compromised in response to B. pertussis. Autopsy studies of infants with fatal pertussis have shown cerebral haemorrhage, oedema (swelling), and cortical atrophy, suggesting that pertussis-related encephalopathy is likely to be associated with pathological changes in the cortex. In the present study we examined the cellular structure and function of the BBB in the cortex during infection of neonatal mice with B. pertussis.

Methods

Ethics approval

Ten-day old male and female C57BL/6J neonatal mice were used in this study. Mice were housed in a pathogen-free facility in the Comparative Medicine Unit, Trinity College Dublin. Animal experiments received local ethical approval (Animal Research Ethics Committee, Trinity College Dublin; Reference number: 060619) and were conducted in accordance with the European Union regulations under project licence (AE19136/P117) authorized by the Irish Health Products Regulatory Authority.

Respiratory challenge with Bordetella pertussis

Litters of ten day old (P10) mice were aerosol infected with B. pertussis (Bp338 strain) from a culture of 1 × 10⁹ CFU/mL using a nebulizer (PARI TurboBOY SX) for 10 min as previously described [8]. Control animals were exposed to water in the aerosol. Live bacteria were enumerated in the lung, nose and brain tissue homogenates by quantify counting colony forming units (CFU) using Bordet-Gengou agar plates for B. pertussis. In all experiments, animals were euthanized 6 days post-aerosol challenge with B. pertussis.

Confocal immunofluorescence staining

Animals were deeply anesthetized with sodium pentobarbital and a transcardial perfusion was performed using 0.89% sterile saline followed by 4% paraformaldehyde (pH 7.4) as fixative. Brains were subsequently post-fixed for 24 h and transferred to 30% sucrose solution for 72 h for cryoprotection. Free-floating immunohistochemistry was conducted on 30 μm thick sections of the piriform cortex that were sectioned using a cryostat. Brain sections were washed in PBS, permeabilized using 0.3% Triton X-100 and blocked using either 10% normal goat serum or 10% normal horse serum for 45 min. All primary antibody incubations were conducted overnight at 4 °C. The antibodies used in the studies are detailed in Table 1. For vessel-associated microglia or astrocytes, sections were incubated with either goat-anti-Iba1 (1:500; ab5076; Abcam) or chicken anti-GFAP (1:2000; ab4674; Abcam) respectively and rabbit anti-AQP4 (1:1000; Abcam; ab282586). For S100β⁺ astrocytes, sections were incubated with rabbit monoclonal antibody [EP1576Y] to S100β (1:500; ab52642; Abcam). For vessel-associated phagocyte analysis, sections were incubated with rabbit anti-CD68 (1:200; ab125212; Abcam). For pericyte and endothelial cell labelling, sections were incubated with rabbit anti-PDGFR-β (1:1000; ab32570; Abcam) and rat anti-CD31 (1:500; ab7388; Abcam) respectively. For erythrocyte labelling, sections were incubated with rabbit anti-myeloperoxidase (1:500; ab221847; Abcam). For endothelial adherens junction labelling, sections were incubated with rabbit anti-VE-cadherin (1:500; ab205336; Abcam). Sections were washed three times in PBS before incubation in secondary antibodies conjugated to AlexaFluor fluorochromes containing 1% BSA in PBS for 2 h in the dark (i.e. Alexa fluor 488 or Alexa fluor 647). Sections were washed in PBS, counterstained with DAPI, washed again and finally mounted onto gelatin-coated slides using aqueous fluoroshield mounting media with DAPI (ab104139; Abcam). On average, 12 images of the piriform cortex per animal were quantified per immunohistochemistry assay performed.

Table 1.

Antibodies used to detect cell types

Cell type	Primary antibody	Secondary antibody
Microglia	Goat IBA1 (1:500; ab5076; Abcam) & Rabbit CD68 (1:200; ab125212; Abcam)	Anti-Goat Alexa Fluor 488 & Anti-Rabbit Alexa Fluor 647
Astrocytes	Chicken GFAP (1:2000; ab4674; Abcam), Rabbit S100β (1:500; ab52642; Abcam) and Rabbit anti-AQP4 (1:1000; Abcam; ab282586).	Anti-Chicken Alexa Fluor 488, Anti-Rabbit Alexa Fluor 647 & Anti-Rabbit Alexa Fluor 647
Pericytes	Rabbit PDGFR-β (1:1000; ab32570; Abcam)	Anti-Rabbit Alexa Fluor 647
Endothelial cells	Rat anti-CD31 (1:500; ab7388; Abcam)	Anti-Rat Alexa Fluor 488
Erythrocytes	Rabbit anti-Myeloperoxidase (1:500; ab221847; Abcam).	Anti-Rabbit 647

Confocal microscopy, image acquisition and analysis

Confocal microscopy was performed using a Leica SP8 STED confocal microscope (x40 and ×63 objective lens with oil immersion). 70–85 z-stack planes were taken with 0.30 µm spacing and images were processed on the LAS X Life Science Microscope Software Platform. Morphometric cell counting of microglia, astrocytes and pericytes and assessment of vasodilation was performed using the ‘’Cell Counter’’ plugin and ‘’Shape Descriptors’’ plugin and ‘’Analyze skeleton’’ plugin on FIJI. For blood vessel diameter analysis, images were converted to 8-bit, a threshold was set, and binary images were then pixel filtered under the ‘’Analyze particles’’ function to reduce background noise. A mask was applied to accurately define the vessel edges and the Feret’s diameter tool was applied to measure the mean diameter (µm). For assessment of pericyte coverage of cerebral blood vessels, Imaris software was used for 3D volume rendering of the CD31⁺ vascular lumen and PDGFR-β⁺ pericytes. After background subtraction using median and Gaussian filters, CD31⁺ endothelial cells and PDGFR-β⁺ pericytes and nuclei were surface rendered with 0.3 μm, 0.2 μm & 0.5 μm smoothing respectively. The area of PDGFR-β directly covering the CD31⁺ vascular lumen was expressed as a percentage of total CD31⁺ endothelial area per image. CD31 is a marker for all vascular endothelial cells including lining capillaries. We did not distinguish between capillaries and larger blood vessels in the present study. To quantify vessel-associated microglia or astrocytes, the total number of Iba1⁺ DAPI⁺ cells (for microglia) or GFAP⁺ DAPI⁺ cells (for astrocytes) were quantified per cortical field and the proportion of microglia or astrocytes colocalising with AQP4 + astrocytic end-feet were enumerated per image and expressed as a % of the total cell count. To assess microglial engulfment of AQP4, background subtraction using median and Gaussian filters was conducted using Imaris (Bitplane). Iba1⁺ cells and AQP4⁺ astrocytic end feet were surface rendered with 0.2 μm and 0.1 μm smoothing, respectively. A mask was applied within Iba1⁺ reconstructed microglia for AQP4 and the volume of engulfed AQP4 within microglia was then measured.

RT-qPCR

RNA was isolated from the cortex using a Nucleospin RNAII Kit (Macherey-Nagel, Duren, Germany) and cDNA was prepared using a High-Capacity cDNA RT kit according to the manufacturer’s instructions (Applied Biosystems, UK). Real-time PCR was performed with predesigned Taqman gene expression assays using an Applied Biosystems 7500 Fast Real-Time PCR machine (Applied Biosystems, Germany) with β-actin as the endogenous control to normalize gene expression data as previously described [9]. mRNA gene transcript expression for gfap was calculated relative to the endogenous control samples (β-actin) and to the experimental control (vehicle) sample giving an RQ value (2^− DDCt), where Ct is the threshold cycle.

Assessment of activated microglia by flow cytometry

Mice were sacrificed and perfused with 20 ml of ice-cold PBS (pH 7.4) before removal of the brain. The brains were collected in complete RPMI medium and chemically digested using collagenase-D (1 mg/ml; Roche) and DNase I (10 mg/ml; Sigma-Aldrich) for 1 h at 37 °C with agitation. Brain tissue homogenates were resuspended in 40% isotonic Percoll and layered over 70% isotonic Percoll before centrifugation (1600 rpm for 20 min). Mononuclear cells were removed from the interface of the Percoll gradients, passed through a 70 μm filter and washed in complete RPMI medium. Ammonium chloride-potassium (ACK) buffer was used to lyse red blood cells. The cells were washed in PBS and incubated in LIVE/DEAD Aqua (1:600; Invitrogen) followed by Fcγ block (αCD16/CD32 FcγRIII; 1:200; BD Biosciences) to block IgG Fc receptors. Cells were then washed and surface-stained with antibodies specific for CD11b (clone: M1/70) and MHC-II (M5/114.15.2),

Statistical analysis

Statistical analysis was performed using Prism 9 (GraphPad Software). Data were analysed using Student’s unpaired 2-tailed t-test on parametric data for independent means. Each data point on the histograms represents data from an individual animal. Data are expressed as the mean ± standard error of the mean (± SEM). Values of p < 0.05 were considered statistically significant.

Results

Infection of neonatal mice with B. pertussis leads to dissemination of bacteria to the brain and alterations in endothelial cell tight-junction integrity

We first confirmed that neonatal respiratory infection with B. pertussis disseminates beyond the airways into the brain. Neonatal (p10) mice were exposed by aerosol to a culture of 1 × 10⁹ CFU/ml. Assessment of CFU counts in tissues at day 6 post challenge revealed a high bacterial load in the lung and nose, as well as disseminating bacteria from the respiratory tract to the brain (Fig. 1A).

Fig. 1.

Cerebral vasodilation and endothelial tight junction integrity during infection of neonatal mice with B. pertussis. Neonatal (P10) mice were aerosol challenged with B. pertussis and B. pertussis infected (Bp) and uninfected control (Con) mice were euthanized 6 days later. A B. pertussis CFU in lung, nose and brain. B Quantification of PDGFR-β⁺ pericytes in cortex. C Vascular PDGFR-β⁺ pericyte coverage. D PDGFR-β + pericyte perimeter length (µm). E Vascular CD31⁺ endothelial cell area. F Representative confocal z-stacks of CD31⁺ endothelial cells (green), PDGFR-β⁺ pericytes (red) and DAPI⁺ nuclei (blue). G Imaris reconstruction of PDGFR-β⁺ pericytes (magenta) CD31⁺ endothelial cells (green) and DAPI⁺ cell nuclei (blue). H Blood vessel diameter in cortex at day 6 post aerosol challenge. I Blood vessel erythrocytes (red) at day 6 post aerosol challenge. J VE-cadherin in cortex 6 days post aerosol challenge with B. pertussis. K Representative confocal images of VE-cadherin (yellow) and DAPI (blue). Each data point on the histograms represents an individual animal. Scale bar in (F) = 200 μm, in (G) = 10 μm and in K = 100 μm. Data are presented as means ± SEM (n = 8). *P < 0.05 **P < 0.01 ***P < 0.001 vs. control by Student’s unpaired Two-tailed t-test

Brain pericytes play a critical role in regulating the BBB by maintaining endothelial cell tight junctions, polarizing astrocyte end-feet and regulating cerebral blood flow by wrapping around capillaries to provide structural support [10]. We used PDGFR-β as a marker of pericytes and found that cell counts were similar in cortical tissue of B. pertussis-infected and uninfected mice (Fig. 1B). However, we observed a significant reduction in the coverage of PDGFR-β⁺ pericytes around CD31⁺ endothelial cells in cortical blood vessels of neonatal mice infected with B. pertussis compared with uninfected mice (p < 0.05, Fig. 1C). Pericyte cell perimeter length was reduced (p < 0.01, Fig. 1D), while cerebral vascular endothelial cell area and blood vessel diameter were enlarged in the cortex of mice infected with B. pertussis (p < 0.001, Fig. 1E and H; see Fig. 1F, G, I). Furthermore, given the critical role of adherens junction proteins in regulating vascular permeability we assessed endothelial cell-specific VE-cadherin signal intensity in the cortex. VE-cadherin was significantly reduced in response to neonatal infection with B. pertussis (p < 0.05, Fig. 1J + K). The reduction in VE-cadherin suggests that infection increases BBB permeability.

Infection of neonatal mice with B. pertussis induces a reactive astrocyte phenotype at the BBB

Astrocytes, and particularly astrocytic end feet, play a critical role in maintaining the BBB. We first assessed cortical astrocyte reactivity using the archetypal marker, GFAP, and found that there was an increase in the number of GFAP⁺ blood vessel-associated astrocytes in cortical tissue at day 6 post challenge with B. pertussis (p < 0.001, Fig. 2A). In addition, gfap mRNA was significantly increased in cortical astrocytes from infected, compared with uninfected, mice (p < 0.05, Fig. 2B). Expression of AQP4, a water channel expressed on astrocytic end feet, was reduced in GFAP⁺ blood vessel-associated astrocytes of infected mice (p < 0.05, Fig. 2C-E). We analysed S100β as a further marker of BBB disruption [11] and observed an increased number of S100β⁺GFAP⁺ astrocytes in the cortical parenchyma of infected mice (p < 0.001, Fig. 2F) and elevated vascular deposition of S100β in blood vessels (Fig. 2G).

Fig. 2.

Reactive astrocytes during infection of neonatal mice with B. pertussis. Neonatal mice were aerosol challenged with B. pertussis and euthanized 6 days later along with un-infected control mice. A Quantification of vessel-associated GFAP⁺ astrocytes per cortical field. B gfap mRNA expression in cortex at day 6 post aerosol challenge. C AQP4 expression on GFAP⁺ astrocytes in cortex. D Representative confocal images of astrocytes (GFAP (gold), AQP4 (red) and DAPI (cyan)) in cortex at day 6 post aerosol challenge. E High magnification 3D render of vessel-associated GFAP⁺ astrocytes in cortex. F S100β + vessel-associated GFAP⁺ astrocytes in cortex at day 6 post aerosol challenge. G Representative confocal images of astrocytes (GFAP (green), S100β (red) and DAPI (blue)). Each data point in the histograms represents an individual animal. Scale bar = 200 μm in (D). Scale bar in (E) = 10 μm and in (G) = 200 μm. Data are presented as means ± SEM (n = 5–16). *P < 0.05, ***P < 0.001 vs. control by Student’s unpaired Two-tailed t-test

Infection of neonatal mice with B. pertussis leads to activation of blood vessel-associated microglia

When compared with uninfected control mice, Iba1⁺ microglia in the cortex of B. pertussis infected mice had the appearance of reactive cells with processes that were markedly thicker and the area occupied by cells was significantly greater (p < 0.001, Fig. 3A). Morphologically, microglia from infected mice had lower voxel counts and fewer process endpoints per cell, indicative of microglia with a larger cell soma with fewer projections (p < 0.001, Fig. 3B-C). A subset of these microglia were in direct proximity to AQP4⁺ astrocytic end-feet, indicating physical interaction with blood vessels. The frequency of these vessel-associated microglia increased in neonatal mice infected with B. pertussis (p < 0.001, Fig. 3D). Based on their reactive cellular morphology and proximity to blood vessels, we hypothesized that reactive microglia may migrate to blood vessels and phagocytose AQP4⁺ astrocyte end-feet in response to infection with B. pertussis. Consistent with this, we found that the frequency of vessel-associated microglia with AQP4⁺ inclusions was significantly increased in neonatal mice infected with B. pertussis (p < 0.001, Fig. 3E-G). We observed greater CD68⁺ lysosomes in the vicinity of the CD31⁺ vascular lumen of the cortex of infected compared with control mice (p < 0.0001, Fig. 3H + I). Furthermore, we recorded a higher number of MHCII⁺ activated microglia in the brains of infected mice by flow cytometry (p < 0.05, Fig. 3J). These findings suggest that Infection of neonatal mice with B. pertussis increases microglial activation and specifically enhances blood vessel-associated activated microglia, which facilitates engulfment of astroglial end-feet.

Fig. 3.

Blood vessel-associated microglia disrupt the BBB during infection of neonatal mice with B. pertussis. Neonatal mice were aerosol challenged with B. pertussis and euthanized 6 days later along with un-infected control mice. A Iba1⁺ microglia coverage in the cortex. B Number of voxels per microglia, indicative of morphological activation. C Endpoints per Iba1⁺ microglia. D Quantification of vessel-associated microglia in cortex. E Quantification of vessel-associated microglia with AQP4 inclusions. F Representative confocal images of Iba1 (green), AQP4 (red) and DAPI (blue) in cortex showing a subset of microglia (insert) in direct proximity to AQP4 + astrocytic end-feet, indicating physical interaction with cortical blood vessels. G Imaris 3D reconstruction of vessel-associated microglia (cyan), AQP4 (red), internalised AQP4 (yellow) and nuclei (magenta). H Quantification of CD68⁺ vessel-associated lysosomes I Representative images of CD31 (green) CD68 (red) and DAPI (blue) staining in cortex. J Quantification of MHCII⁺ microglia by flow cytometry with representative FACS plot. In the histograms, each point represents data from an individual animal, except for (B) & (C) where data points represent an individual cell. Scale bar = 100 μm in (F). Scale bar in (G) = 10 μm and in (I) = 50 μm. Data are presented as means ± SEM (n = 8).***P < 0.001 **** P < 0.0001 vs. control by Student’s unpaired Two-tailed t-test

Discussion

The findings of this study demonstrate that infection of neonatal mice with B. pertussis leads to disruption of the BBB and migration of reactive microglia to cerebral blood vessels. These microglia were increased in number and had several morphological features consistent with a reactive state. They were also phagocytic and internalised a large volume of AQP4, a marker of astrocytic end-feet that are a major component of the BBB. Therefore, the study has uncovered a link between blood vessel-associated microglia and vascular permeability during early life infection of mice with B. pertussis. Reactive, vessel-associated microglia, have previously been shown to phagocytose BBB components in response to chronic systemic inflammation in a mouse model of lupus, and these findings were also associated with leakage of blood components and widespread neuroinflammation [12].

Cortical astrocytic end-feet are a critical component of the BBB and aid barrier function by controlling local blood flow via the release of vasoactive molecules and calcium transients that mediate vasodilation and vasoconstriction [13]. Despite an increase in the proportion of GFAP-enriched, vessel-associated cortical astrocytes in response to infection of neonatal mice with B. pertussis, we observed a reduction in AQP4 expression on end-feet projections enwrapping the blood vessel endothelium. Glial-specific loss of AQP4 reduces water uptake [14], whereas increased AQP4 activation improves BBB integrity and perihematomal oedema after intracerebral haemorrhage [15, 16]. It is possible that the perivascular loss of AQP4 on astrocytic end-feet in infected mice reported here is attributable to the increased phagocytosis by vessel-associated microglia, which has previously been shown to be triggered by systemic inflammation [12].

Circulating astrocyte-derived S100β is considered to be a biomarker of BBB permeability [17]. We found increased numbers of vessel-associated S100β⁺ cortical astrocytes and enhanced vascular deposition of S100β in neonatal mice infected with B. pertussis. S100β is also a secreted alarmin that can act via the RAGE/TNF-α pathway on microglia to stimulate cyclooxygenase-2 expression and production of the potent vasodilator prostaglandin E₂ [18]. MRI on patients with severe pertussis in infancy often present with encephalitis and aggravated, diffuse brain swelling due to cerebral vasodilation, findings which are accompanied by marked increases in inflammatory cytokines in CSF [2]. Given the potent role of astrocyte-derived S100β in activating microglia, it is a likely driver of neuroinflammation in our model. Other groups have shown that S100β promotes microglial activation in rodent models of neurodegeneration, and that its ablation results in reduced microgliosis and protects against chronic EAE pathology in mice [19, 20].

We assessed the impact of infection of neonatal mice with B. pertussis on brain pericytes because they are critical for regulating cerebral blood flow and the integration of endothelial cell and astrocyte functions at the neurovascular unit [10]. We found changes in vascular endothelium pericyte coverage during infection of neonatal mice with B. pertussis, suggesting dysregulation in cortical blood flow due to deficits in pericyte-endothelial cell interactions. This was associated with reduction in pericyte cell perimeter length, indicating morphological alterations in brain pericytes post-infection. Furthermore, CD31⁺ endothelial cell area and blood vessel diameter were increased, indicating cerebral vasodilation in infected mice. It is possible that the increased vasodilation that occurs in the brain during infection, in conjunction with the reduction in endothelial cells pericyte coverage, will impact cerebral blood flow. This is consistent with CNS vasculature complications observed during critical pertussis in human infancy. Significantly, pericyte loss can cause BBB breakdown and hypoperfusion due to defective vasoregulation and vascular regression, preceding the onset of severe cognitive deficits [21]. Indeed, we have previously shown that peripheral immune cells, including neutrophils, monocytes and T-cells, infiltrated the brain at day 6 post neonatal challenge with B. pertussis. This, and our observation that fibrinogen, a high molecular weight protein, was present in the brain parenchyma at the same time, indicates that the BBB is more permeable during B. pertussis infection [7]. In addition to these changes, we found that VE-cadherin was decreased during infection. VE-cadherin plays a critical role in blood vessel formation and in regulating vascular permeability by maintaining endothelial cell junction integrity [22]. The infection-related decrease in VE-cadherin is consistent with destabilisation of brain microvascular endothelial barrier function under pathological conditions [23].

Severe early life infections are increasingly associated with neurodevelopmental disorders, including autism spectrum disorder (ASD) [24]. Our findings of BBB disruption and vessel-associated glial reactivity during this critical window in early postnatal brain development may be a contributing factor toward lasting neural circuit alterations and the onset of ASD-related behaviour.

Study limitations

There are limitations to this study. Whether the BBB alterations reported in our study precede, coincide with, or follow microglial and astrocytic reactivity is currently unknown. We have not determined whether vessel-associated glial reactivity triggers BBB disruption during neonatal B. pertussis infection or occurs because of enhanced vascular permeability. We focussed on one timepoint (day 6) post challenge with B. pertussis due to the critical role microglia play in shaping brain development around this time. Studies on earlier and later timepoints should provide further insight into the mechanisms underpinning BBB disruption in response to neonatal infection.

In conclusion, we report a link between blood vessel-associated reactive microglia and BBB disruption during infection of neonatal mice with B. pertussis and propose that phagocytosis of astrocytic end-feet may be a key factor in this disruption. These findings were associated with evidence of cerebral vasodilation, malformations in brain pericytes and alterations in adherens junction proteins, which collectively contribute to a leaky BBB. Our work has significant implications for our understanding of the acute neurological complications and long-term neurological sequelae associated with critical pertussis in human infants.

Abbreviations

AQP4

Aquaporin-4

BBB

Blood–brain barrier

CD31

Cluster of Differentiation 31

Cdh5

vascular endothelial cadherin (VE-cadherin)

CNS

Central nervous system

CSF

Cerebrospinal fluid

EAE

Experimental autoimmune encephalomyelitis

EC

Endothelial cells

IL-6

Interleukin-6

IL-10

Interleukin-10

MBP

Myelin basic protein

PBS

Phosphate Buffered Saline

PGE₂

Prostaglandin E2

S100β

S100 calcium-binding protein B

TNF-α

Tumor necrosis factor-alpha

VAM

Vessel-associated microglia

Authors’ contributions

Study design: E.O.N, K.H.G.M., M.A.L.; supervision and funding acquisition: K.H.G.M., M.A.L. and E.O.N.; methodology, data collection and visualization: E.O.N.; writing, review and editing article: E.O.N., K.H.G.M., M.A.L.

Funding

This work was supported by research grants from Science Foundation Ireland / Taighde Éireann – Research Ireland under Grants number 16/IA/4468 and 22/FFP-A/10297 to K.H.G.M and from The Trinity Trust to E.O.N.

Data availability

Data supporting the findings of this study are available within the paper. Raw datasets are available from the corresponding author on reasonable request.

Declarations

Ethical approval and Consent to participate

Not applicable.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.