Background
Bacterial meningitis is a devastating disease defined by meningeal inflammation in response to bacterial infection [
1,
2]. Despite vaccination efforts, the human-specific, Gram-negative bacterium
Neisseria meningitidis (Nm, meningococcus) remains one of the leading causes of bacterial meningitis worldwide [
3‐
5]. The pathogen asymptomatically colonizes the nasopharynx of up to 35% of the healthy population, depending on age and other risk factors [
6].
N. meningitidis can cross the epithelial nasopharyngeal barrier, disseminate systemically via the bloodstream, and cause invasive meningococcal disease (IMD), such as septicemia and/or meningitis [
7]. Although treatable with modern antibiotic therapy, systemic meningococcal infection is still associated with neurological sequalae and mortality [
1,
2].
A critical step in the pathogenesis of meningococcal meningitis is the traversal of the meningeal blood-cerebrospinal fluid barrier (mBCFSB) and subsequent interaction with leptomeningeal cells (LMCs) [
8,
9]. Blood-central nervous system (CNS) barriers such as the blood–brain barrier (BBB) and the mBCSFB are comprised of highly specialized brain endothelial cells (BECs) that maintain CNS homeostasis by facilitating transport of nutrients and restricting passage of toxins, drugs, and pathogens [
1]. One of the major phenotypes unique to BECs compared to peripheral ECs is the presence of complex adherens and tight junctions that minimize space between adjacent cellular membranes and prevent paracellular passing of material into the CNS [
10,
11]. BECs of the mBCSFB make up blood vessels that run through CSF-filled spaces between the meninges, for instance the subarachnoid space [
8,
9,
12]. In contrast to the BBB where BECs are surrounded by cells of the neurovascular unit, namely astrocytes and pericytes [
10,
11], BECs of the mBCSFB are surrounded by and enclosed in sheets of LMCs that make up the arachnoid and the pia mater [
8].
N. meningitidis interaction with BECs has primarily been evaluated using immortalized ECs and BECs in vitro due to the human-specific nature of the pathogen (reviewed in [
13]). Previous studies have identified several virulence factors important for adherence to and invasion of BECs such as type-IV pili (Tfp), the opacity proteins OpcA and Opa as well as a series of minor adhesion or adhesion-like proteins (e.g. adhesin complex protein, Neisserial adhesin A,
Neisseria hia homologue A protein or the autotransporter meningococcal serine protease A). Furthermore, corresponding binding host-cell receptors such as CD147 and α5β1/αvβ3 integrins were discovered [
14‐
22]. Effects on signaling pathways and rearrangement of cytoskeletal as well as cell-membrane and junction components in BECs have been described in detail after infection of these cells with
N. meningitidis [
14‐
16,
18,
23‐
28].
Finally, two major routes for meningococcal traversal of the BEC barrier have been proposed, the first being an intra- or transcellular pathway resulting from tight interaction of bacterial adhesins/ invasins and cellular receptors as well as vesicular uptake after formation of microvillus-like structures around the bacteria [
17,
18,
27,
29‐
32]. Inter- or paracellular crossing by
N. meningitidis has been suggested by several studies reporting that barrier integrity is compromised under bacterial infection due to rearrangement, degradation, or downregulation of cell-junction components [
15,
25,
28,
33].
Relatively little is known about the subsequent meningococcal traversal of the LMC sheet enclosing the vessel endothelium and
N. meningitidis interaction with LMCs of the arachnoid and pia mater. Previous studies have identified bacterial factors that contribute to
N. meningitidis-LMC interaction such as Tfp and have characterized the inflammatory response of LMCs to infection [
34‐
37]. In these studies, LMCs derived from meningiomas were used, which have been validated for infection studies with
N. meningitidis and other bacterial pathogens of the CNS [
8,
34‐
39].
Up to this point, studies of
N. meningitidis interaction at the mBCSFB have primarily utilized monocultures of immortalized BECs, which do not retain critical BBB phenotypes such as high barrier tightness [
11,
40,
41]. Although some studies show that co-culture with other cell types of the CNS was able to improve barrier phenotype, defining measurables such as transendothelial electrical resistance (TEER) remained relatively low [
11,
40,
41]. The immortalized microvascular endothelial cell line hCMEC/D3 is a robust and widely utilized in vitro model to study
N. meningitidis-BEC interaction, although can lack some key BEC phenotypes [
14‐
16,
42,
43].
Recent advances in stem-cell technologies have generated model BEC-like cells derived from human induced pluripotent stem cells (iPSCs) that better reflect the barrier phenotype of BECs [
11,
44,
45]. iPSC derived BECs (iBECs) show characteristic expression of BBB markers such as adherens and tight junction components, exhibit high TEER and respond to co-culture with other cell types of the CNS [
11,
40,
44‐
49]. Recently, iBEC monoculture models have been validated for infection studies with meningeal pathogens such as
Streptococcus agalactiae (group B streptococcus) [
50‐
52], and we have begun to use iBECs monoculture models to study their interaction with
N. meningitidis [
33,
46]. Additionally, this model has also been utilized to examine viral pathogens with neurotropism [
53,
54]. However, to our knowledge, effects of co-culture with other CNS cell types on host–pathogen interaction at the mBCSFB has not yet been evaluated.
Here, we report on the development of a physiologically relevant in vitro model of the human mBCSFB using BECs derived from induced pluripotent stem cells in co-culture with meningioma-derived LMCs to examine N. meningitidis interaction. In parallel, we developed BEC-LMC co-culture models using the established infection model cell line hCMEC/D3 for reference.
Discussion
Bacterial meningitis is a severe disease that occurs when pathogens such as
Neisseria meningitidis (the meningococcus) cross the meningeal blood-cerebrospinal fluid barrier (mBCSFB) and infect the meninges [
1,
2]. The mBCSFB consists of specialized brain endothelial cells (BECs) that exhibit a barrier phenotype to maintain brain homeostasis and are surrounded by leptomeningeal cells (LMCs) [
1,
8]. Due to the human-exclusive tropism of
N. meningitidis, most studies examining meningococcal interaction at the mBCSFB have utilized primary or immortalized BECs that, however, lack critical barrier phenotypes in vitro [
11,
40,
41].
N. meningitidis interaction with LMCs alone has been investigated using LMCs derived from meningioma [
34‐
37]. However, meningococcal penetration of the mBCSFB has not yet been studied in a multicellular context including this cell type in vitro. Here, we used iPSC derived BECs (iBECs) or hCMEC/D3 cells in co-culture with meningioma-derived LMCs to develop a more complex and physiologically relevant in vitro model for studying
N. meningitidis-mBCSFB interaction.
Hallmark phenotypes of BECs include endothelial markers, tight junction expression, barrier properties, functional nutrient and efflux transporters, and response to other CNS cell types [
74]. To benchmark and validate BEC in vitro models, certain methods including TEER measurements, permeability assays, and immunostainings of key markers such as endothelial adherens and tight junctions are used [
40]. Primary and immortalized BECs are scalable and have been widely used for modeling of blood-CNS barriers but frequently lose important barrier phenotypes once removed from their native microenvironment [
11]. The extensively characterized immortalized microvascular endothelial cell line hCMEC/D3 retains many BEC characteristics but exhibits low TEER and often lacks continuous expression of tight junction components at the cell–cell junctions [
11,
41‐
43,
75] (Additional file
2: Fig. S2). Despite these drawbacks, hCMEC/D3s are a robust and widely used in vitro model to study
N. meningitidis-BEC interaction [
14‐
16,
28].
Advances in stem-cell technologies have generated model brain endothelial-like cells derived from iPSCs, which possess all relevant BEC phenotypes including endothelial markers, tight junction expression, barrier properties, response to other CNS cell types, and functional efflux transporters [
11,
44,
45,
48,
49,
66]. Following the publication of the initial protocols, much research has been conducted on these models, uncovering advantages as well as weaknesses, and establishing alternative protocols and improvements. Described limitations are the expression of epithelial genes and proteins, which has been detected in addition to the endothelial phenotypes described [
76]. In this study, we differentiated iBECs from iPSC line IMR90-4 according to previously published protocols [
44,
46,
48] and co-cultured them with LMCs on permeable transwell inserts. We observed clear expression of important brain endothelial markers such as CD31, VE-cadherin, and claudin-5 within the iBEC layer, but epithelial characteristics were also present, particularly related to the cytoarchitecture, such as small protrusions on the apical surface, larger cell height and nucleus, and epithelial-like organization of intercellular junctions (Fig.
2). Although these limitations must be considered when employing the model, it remains suitable for certain applications due to its main advantages such as the tight barrier properties. Recently, iBECs have been useful for modeling various diseases of the CNS including Huntington’s disease, MCT8 deficiency (causing Allan-Hurndon-Dudley syndrome), and infectious disease [
33,
46,
50‐
53,
77‐
79].
Important benchmarks for the barrier properties of in vitro BBB models are high TEER inversely related to paracellular permeability of solutes (although non-linear) [
40], although in vivo data are are only available for pial microvessels in anesthetized frogs and rats [
80]. This relationship (one-phase exponential decay) was first demonstrated on rat primary BECs, where permeability coefficient values for sodium fluorescein were below 2 × 10
–6 cm/s above a threshold TEER of 130 Ω cm
2 [
81]. iPSC derived BECs typically reach TEER values above 1500 Ω x cm
2 and NaF permeability values in the order of 10
–7 cm/s [
33,
44,
46,
47,
49,
66] (Fig.
2c,d). In this study, we observed that LMC co-culture further increased iBEC barrier tightness and stability over multiple days as reflected by lower NaF permeability and higher TEER. While iBECs alone can exhibit high paracellular tightness, co-culture with other cells from the neurovascular unit such as astrocytes and pericytes has been reported to assert stimulating as well as stabilizing effects on iBEC barrier properties, demonstrating that iBECs respond to cues from other CNS cell types [
11,
40,
44,
45,
49,
66,
82‐
84]. Astrocytes and pericytes have also been shown to induce BBB properties in primary bovine, porcine, rodent, or primate BEC models, which exhibited physiologically relevant levels of paracellular restriction that human primary and immortalized BEC lines did not reach [
11,
40,
41,
85]. Co-culture with leptomeningeal cells, which are important in the context of the human mBCSFB, has not been explored before. We observed slightly higher TEER and lower NaF permeability of hCMEC/D3 cells co-cultured directly on the transwell membrane with LMCs compared to monoculture, although these effects could not be correlated with changes in barrier phenotype of hCMEC/D3 layers due to the low overall TEER.
Recognizing the advantage of a human in vitro system exhibiting physiological barrier tightness, an increasing number of studies has recently used iPSC derived BECs to model interaction with CNS pathogens such as GBS, Zika virus, and more recently SARS-CoV2, particularly to investigate how such pathogens affect and penetrate the blood-CNS barriers [
33,
46,
50‐
54]. Recently, we validated iPSC derived BECs for infection studies with the human-specific bacterium
N. meningitidis [
33]. As monoculture in vitro models only distantly represent the native microenvironment and the use of in vivo models to study interaction with human-specific
N. meningitidis is limited to humanized rodents [
86], more complex in vitro models could be useful to study this host–pathogen interaction. Co-culture systems with iBECs and other CNS cell types are now widely used to model function and dysfunction of blood-CNS barriers [
40,
44,
45,
49,
66,
82‐
84] but have not been used for infection studies with CNS pathogens yet. Therefore, we developed and used the iBEC-LMC co-culture model to examine meningococcal interaction with and traversal of the mBCSFB.
We observe substantial bacterial adherence to iBECs in our iBEC-LMC co-culture system soon after infection, consistent with our results using hCMEC/D3s with LMC co-culture as well as published data [
14,
17,
18,
26]. This tight interaction, which is primarily mediated by meningococcal type IV pili (Tfp), is critical for vascular colonization and, ultimately, penetration of the mBCSFB [
14‐
18,
26,
28]. The mechanism of barrier traversal by
N. meningitidis has been a matter of debate for a long time. Most bacteria that can cause meningitis including
N. meningitidis, Group B
Streptococcus,
Streptococcus pneumoniae, and
Escherichia coli K1, penetrate blood-CNS barriers such as the mBCSFB via a transcellular pathway following cellular invasion by the bacteria or via a paracellular pathway that becomes available through disruption of cellular junctions or cell damage [
1,
2]. Cellular invasion of BECs by
N. meningitidis has been observed using peripheral, bone marrow derived, brain microvessel derived, and, recently, also iPSC derived BECs in vitro, suggesting a transcellular route for
N. meningitidis traversal of the mBCSFB [
17,
18,
26,
27,
29‐
33]. In this study, we also found that
N. meningitidis invades iPSC derived BECs and hCMEC/D3 cells in our newly developed co-culture systems with LMCs. While relatively low at first, bacterial invasion increased significantly during prolonged infection, which was previously reported in another immortalized BEC model and potentially results from the demasking of adhesins and invasions upon downregulation of the polysaccharide capsule [
18]. Co-culture with LMCs did not affect
N. meningitidis adherence or invasion of BECs in our assays. Finally, we observed increasing meningococcal transmigration of our iBEC and iBEC-LMC models within 24 h of infection while barrier integrity was still very high. Considering that these models exhibit physiological barrier tightness, these observations further support the hypothesis of transcellular barrier traversal by
N. meningitidis.
Interestingly, we observed higher rates of bacterial transmigration often correlated with lower TEER. For instance, compared to data from the iBEC models, transmigration rates were already substantially higher in the hCMEC/D3 based models at the earliest measured time point after infection, and we even detected more meningococci traversing the iBEC monoculture than the iBEC-LMC co-culture model early on, although absolute counts were low. This suggests that using models exhibiting physiological barrier tightness may be important for studying bacterial traversal, and it indicates that
N. meningitidis likely crosses the BEC barrier via a paracellular route if available. Previous studies have suggested that this pathway does become available through disruption of cellular junctions [
15,
25,
28,
33]. Increased permeability to lucifer yellow and discontinuous junctional localization of adherens junction protein VE-Cadherin was detected and correlated with meningococcal BEC barrier traversal in infected hCMEC/D3s [
15,
28]. Mechanistically, junctional disorganization was caused by signaling events triggered by Tfp mediated interaction between
N. meningitidis and hCMEC/D3s that lead to the recruitment of cytoskeletal and cell-junction components underneath adherent meningococcal colonies [
15,
28]. Using another immortalized BEC cell line (HBMEC), cleavage of tight junction protein occludin and cell detachment mediated by matrix-metalloproteinase MMP-8 was observed upon prolonged
N. meningitidis challenge [
25]. In this study, we observed loss of TEER and reduced junction coverage of occludin in iBECs co-cultured with LMCs after 24 h of infection. Additionally, previous data from iBEC monocultures grown on plate indicates cleavage of occludin induced by
N. meningitidis infection [
33]. Together, this suggests modulation of occludin in infected iBECs, although further investigation is required to fully elucidate this mechanism.
In addition to disorganization of cell-junction components, we examined effects of
N. meningitidis infection on gene expression of adherens and tight junction proteins in BECs as a potential mechanism for barrier deterioration and found that expression of genes coding for VE-Cadherin, ZO-1 and especially endothelial specific tight junction protein claudin-5 was significantly downregulated in infected iBECs, predominantly after 24 h of infection or later. Simultaneously, Snail-1 (
SNAI1), a transcriptional repressor of tight junctions [
67,
68,
70,
71], previously linked to Group B
Streptococcus,
Streptococcus pneumoniae, and
E. coli K1 induced BBB disruption [
69,
87,
88], was upregulated. These results were consistent between iBEC-LMC co-culture, iBEC monoculture on transwell and iBECs cultured on plate [
33]. Furthermore,
N. meningitidis infection had similar effects on expression of cell-junction genes as well as
SNAI1 in hCMEC/D3s with and without LMC co-culture. Together, these findings suggest downregulation of tight and adherens junction genes in addition to reorganization of junction proteins during
N. meningitidis induced barrier disruption. However, transcriptional repression mediated by Snail1 seems not to be the sole mechanism behind this effect, as indicated by siRNA knockdown experiments. Further investigation is required to elucidate this mechanism. In conclusion, while meningococcal invasion of BECs potentially contributes to early traversal, deterioration of barrier properties may open up a more accessible paracellular route later on during infection.
Meningococcal interaction with the mBCSFB and proliferation in the subarachnoid space evokes a strong inflammatory response that is triggered by immune activation of BECs and LMCs and leads to the influx of leukocytes, primarily neutrophils at first [
1,
8]. Previous studies have shown that Group B
Streptococcus and
N. meningitidis elicit upregulation of neutrophilic chemoattractants in iPSC derived BECs [
33,
50]. Consistent with these findings, we observed transcriptional upregulation of IL-8 (
CXCL8),
CXCL1, CXCL2 and
CCL20 in iBECs cultured on transwell inserts with and without LMCs. This demonstrates that iBECs activate innate immune response mechanism in response to bacterial infection. However, despite the transcriptional upregulation of these factors, secretion of the related proteins was almost undetectable as our study as well as previous investigations have shown [
33,
50]. hCMEC/D3s responded to the infection in similar fashion, although all analyzed gene transcripts were upregulated much more substantially and high levels of cytokine secretion were detected in this model. Further investigation is required to fully elucidate the inflammatory response of BECs to bacterial infection and determine if the low abundance of secreted cytokines observed in iBECs is biologically relevant. Activation of LMCs upon meningococcal interaction following mBCSFB traversal likely contributes to inflammation in the subarachnoid spaces, too, and secretion of various cytokines after
N. meningitidis infection has been demonstrated in meningioma derived LMCs [
34,
35,
37].