Two Strikingly Different Signaling Pathways Are Induced by Meningococcal Type IV Pili on Endothelial and Epithelial Cells

Abstract

Following adhesion on brain microvasculature, Neisseria meningitidis is able to cross the blood-brain barrier (BBB) by recruiting the polarity complex and the cell junction proteins, thus allowing the opening of the paracellular route. This feature is the consequence of the activation by the type IV pili of the β₂-adrenergic receptor/β-arrestin signaling pathway. Here, we have extended this observation to primary peripheral endothelial cells, and we report that the interaction of N. meningitidis with the epithelium is strikingly different. The recruitment of the junctional components by N. meningitidis is indeed restricted to endothelial cell lines, and no alteration of the cell-cell junctions can be seen in epithelial monolayers following meningococcal type IV pilus-mediated colonization. Consistently, the β₂-adrenergic receptor/β-arrestin pathway was not hijacked by bacteria adhering on epithelial cells. In addition, we showed that the consequences of the bacterial signaling on epithelial cells is different from that of endothelial cells, since N. meningitidis-induced signaling which protects the microcolonies from shear stress on endothelial cells is unable to do so on epithelial cells. Finally, we report that the minor pilin PilV, which has been shown to be essential for endothelial cell response, is not a required bacterial determinant to induce an epithelial cell response. These data demonstrate that even though pilus-mediated signaling induces an apparently similar cortical plaque, in epithelial and endothelial cell lineages, the signaling pathways are strikingly different in both models.

INTRODUCTION

Neisseria meningitidis (a meningococcus) is responsible for cerebrospinal meningitis and septicemia with, in some circumstances, purpura fulminans. This Gram-negative bacterium is a commensal of the human nasopharynx that colonizes 10 to 20% of a population (1). In only a small proportion of colonized people does the meningococcus invade the bloodstream and cross the blood-brain barrier (BBB), invading the meninges. In addition, common systemic symptoms during meningococcal disease such as purpura are associated with meningococcal interaction with peripheral endothelial cells. Therefore, during its life cycle N. meningitidis has to interact with epithelial cells to colonize the nasopharynx and with endothelial cells, the latter being associated with the clinical peripheral symptoms and the crossing of the BBB.

N. meningitidis is able to adhere and grow on the apical surfaces of endothelial cells both in vitro and in vivo, thus leading to the formation of bacterial microcolonies (15, 22). This ability mainly relies on type IV pili that are long filamentous elements that extend from the bacterial surface. In addition, N. meningitidis possesses several adhesion molecules, such as Opa and Opc, but in the bloodstream they are hidden underneath a capsule (6, 16, 27, 35). The capsule prevents opsonophagocytosis and killing by the complement and is subsequently required to survive in the extracellular fluids. In addition to promoting the adhesion of bacteria on cells and the formation of microcolonies, type IV pili induce signaling on eukaryotic cells, thus leading to the accumulation under the colony of the molecular linker ezrin, actin, adhesion molecules, and membrane receptors (such as ICAM-1 and CD44) that are clustered in a honeycomb-like structure referred to as “cortical plaque” (3, 8, 17).

In a brain endothelial cell line, the formation of the cortical plaque is a consequence of the recruitment and activation by the type IV pili of the β₂-adrenoceptor. Two components of the pili, the major pilin PilE and the minor pilin PilV, have been shown to recruit the β₂-adrenoceptor. The signaling following the activation and recruitment of the β₂-adrenoceptor activates a β-arrestin/Src tyrosine kinase/cortactin pathway that is involved in the formation of the cortical plaque (3, 19). Another consequence of the formation of the cortical plaque is the recruitment of the polarity complex Par3/Par6/PKCζ, followed by the delocalization of the junctional proteins from the intercellular junctions at the site of the bacterium-cell interaction. This delocalization leads to an opening of the paracellular route by which N. meningitidis may cross the brain endothelium (4). However, the activation of a similar signaling pathway has not yet been reported for peripheral endothelial cells.

On epithelial cells, type IV pilus-mediated adhesion of N. meningitidis is also followed by the formation of a cortical plaque, which is believed to be identical to that observed on endothelial cells. However, several reports have shown that N. meningitidis interaction with epithelial cells does not alter the junctions, and bacteria are believed to cross the epithelium using the transcellular route (18, 23, 31). The mechanism by which bacteria are internalized into eukaryotic cells is not fully understood but seems to involve the secondary adhesion molecules Opa and Opc (33, 35, 36). These data obtained using epithelial cells are not consistent with the data obtained using brain microvascular endothelial cells that show cell-cell junction leakage. These discrepancies suggest that type IV pilus-mediated signaling has consequences for epithelial cells different from those observed with brain endothelial cells. Moreover, it has already been reported that adhesion of the meningococcus to endothelial cells and epithelial cells differed significantly depending on the major pilin subunit expressed by the type IV pili (34, 37).

In this report we extend the initial observation of the delocalization of junctional proteins on brain endothelial cells to primary peripheral endothelial cells and demonstrate that the activation of the β₂-adrenoceptor/β-arrestin pathway is limited to brain and peripheral endothelial cells. These results demonstrate that the two main interactions of N. meningitidis on host cells, responsible for the asymptomatic nasopharyngeal colonization on one hand and the clinical symptoms on the other, are the consequence of two different signaling pathways induced by type IV pili.

MATERIALS AND METHODS

Bacterial strains and infection of monolayers.

N. meningitidis strain 2C43, a piliated capsulated Opc⁻ Opa⁻ variant of serogroup C strain 8013 (20), was used unless otherwise specified. The isogenic capsulated piliated pilV mutant and the isogenic piliated noncapsulated strains expressing Opa and the isogenic noncapsulated pilE mutant expressing Opa were previously described (4, 19). A green fluorescent protein (GFP) expressing N. meningitidis strain 2C43 was obtained by introducing the plasmid pAM239 (15). All strains were grown on GCB-agar plates supplemented with appropriate antibiotics in a 5% CO₂ incubator at 37°C.

The day before infection the cell lines were serum starved in appropriate cell medium containing 0.1% bovine serum albumin (BSA) except for human dermal microvascular endothelial cells (HDMECs) and human umbilical vein endothelial cells (HUVECs). On the day of infection, a suspension of the bacteria from an overnight culture on plate was adjusted to an optical density at 600 nm of 0.05 and incubated for 2 h at 37°C in a prewarmed cell culture medium or starvation medium. The cells were infected with bacteria at a multiplicity of infection (MOI) of 100 for 30 min to allow N. meningitidis adhesion and then washed once with cell culture medium or starvation medium and maintained in appropriate fresh medium for 2 or 5 h.

Cell lines.

The immortalized human brain microvascular endothelial cell line hCMEC/D3, which retains the main characteristics of primary brain endothelial cells, was described previously (39). hCMEC/D3 cells were grown at a density of 2.5 × 10⁴ cells/cm² in flasks coated with 5 μg of rat tail collagen type I (BD Bioscience)/cm² in EBM-2 medium (Lonza) supplemented with 2.5% of fetal bovine serum (FBS; PAA Laboratories), 1% penicillin-streptomycin-amphotericin (PSA; PAA Laboratories), 1.4 μM hydrocortisone, 5 μg of ascorbic acid/ml, 10 mM HEPES, and 1 ng of b-FGF/ml at 37°C in a humidified incubator in 5% CO₂. HDMECs were purchased from Promocell and grown in endothelial cell growth medium MV (Promocell) supplemented with 10% FBS, endothelial cell growth medium supplement mix (Promocell), and 1% PSA. HUVECs were obtained from Promocell and grown in endothelial serum-free medium supplemented with 10% FBS, 0.5 IU of heparin (Sigma)/ml, 40 μg of endothelial cell growth supplement/ml, and 1% PSA.

Human colonic carcinoma-derived cells T84 and Caco2 cells were obtained from the American Type Culture Collection (ATCC; references CCL-48 and HTB-37, respectively). Pharynx carcinoma-derived Detroit-562 and FaDu cells were purchased from the ATCC (references CCL-38 and HTB-43). Endometrium carcinoma Hec1b cells were obtained from the ATCC (HTB-113). All of the epithelial cells but the T84 cells were grown in Dulbecco modified Eagle medium (PAA Laboratories) supplemented with 10% FBS and 1% PSA at 37°C in a humidified incubator in 5% CO₂. T84 cells were grown in Ham F-12 medium (PAA Laboratories) supplemented with 10% FBS and 1% PSA.

Immunofluorescence assays.

For the immunofluorescence assays, the cells were grown on glass coverslips coated with 5 μg of rat tail collagen type I (BD Biosciences)/cm² until confluence. For assays with HUVECs, glass coverslips were coated with human fibronectin (10 μg/ml in phosphate-buffered saline [PBS]; Invitrogen). For the transwell assays, the cells were seeded on collagen type I 0.45-μm inserts (BD Biosciences) until confluence. After infection, the cells were fixed using 4% paraformaldehyde for 20 min, washed three times with PBS, permeabilized for 5 min in PBS containing 0.1% Triton X-100, and then washed three times with PBS. After a 20-min incubation in PBS containing 0.3% BSA, the cells were incubated with primary antibodies at the recommended concentration in 0.3% PBS-BSA for 1 h. After three washes with PBS, the cells were incubated with secondary antibodies conjugated to Alexa fluorochrome. DNA was stained using DAPI (4′,6′-diamidino-2-phenylindole) at 0.5 μg/ml. When needed, actin staining was performed with an Alexa-conjugated phalloidin (Invitrogen). After several additional washings, the coverslips were mounted in Mowiol (Biovalley). For three-dimensional (3D) reconstruction assays, image acquisition was performed on a laser-scanning confocal microscope (Leica TCS SP5). Images were collected and processed using Leica Application Suite AF Lite software. 3D reconstruction was performed using IMARIS software.

Quantitative analysis of ezrin (and other proteins) recruitment under bacterial colonies.

The recruitment efficiency was estimated by determining the proportion of colonies that efficiently recruited the protein of interest in a “honeycomb shape” just under the colonies. At least 50 colonies were scrutinized per coverslip. Each experiment was performed in duplicate and repeated at least three times. The data were examined for significance using the Student t test.

Antibodies and drugs.

The following antibodies were used: rabbit polyclonal anti-ezrin (kindly provided by P. Mangeat, CNRS, UMR5539, Montpellier, France), mouse monoclonal anti-claudin-5 antibody (clone 4C3C2; Zymed-Invitrogen), rabbit polyclonal anti-claudin-3 (Zymed-Invitrogen), rabbit polyclonal anti-occludin (Zymed-Invitrogen), rabbit polyclonal anti-ZO1 (Zymed-Invitrogen), mouse monoclonal anti-E-cadherin (clone HECD-1; AbCam), rabbit polyclonal anti-VE-cadherin (Bender MedSystems), mouse monoclonal anti-p120-catenin (clone 98/pp120; BD Transduction Laboratories), mouse monoclonal anti-myc tag (clone 9B11; Cell Signaling), rabbit monoclonal anti-β-arrestin-1/2 (clone D24H9; Cell Signaling), and rabbit polyclonal anti-actin (Sigma-Aldrich). The following goat secondary antibodies were used for immunofluorescence labeling or immunoblotting: anti-mouse IgG(H+L) and anti-rabbit IgG(H+L) coupled to horseradish peroxidase (Jackson Immunoresearch Laboratories) and anti-mouse IgG(H+L) and anti-rabbit IgG(H+L) coupled to Alexa 488, Alexa 546, or Alexa-633 (Invitrogen). Isoproterenol hydrochloride was purchased from Sigma and used at a final concentration of 10 μM (saturating concentration, 10 × K_D). The Src inhibitors PP1 and PP2 were purchased from Sigma and used at final concentrations of 50 and 20 μM, respectively.

Plasmids and transfection assays.

Plasmids encoding Par3-myc and Par6-YFP were generously provided by Ian Macara (University of Virginia at Charlottesville Center for Cell Signaling and Department of Microbiology). YFP, β₂-adrenergic receptor-YFP, β-arrestin1-myc, and β-arrestin2-myc coding plasmids were as described previously (3, 28). 4×CRE-luciferase reporter plasmid was purchased from Stratagene. All of the endothelial cell lines were transfected by using the Amaxa nucleofactor kit for HUVECs (Lonza) according to the manufacturer's instructions. All of the epithelial cell lines were transfected using Lipofectamine 2000 (Invitrogen) according to the manufacturer's instructions.

CRE-luciferase experiments.

The cells were transfected with the 4×CRE-luciferase plasmid (Stratagene) the day before the experiment using a Lipofectamine 2000 kit (Invitrogen). After 2 h of incubation with 10 μM isoproterenol (10 × K_D), a specific β₂-adrenergic agonist, the cells were washed twice with ice-cold PBS and lysed using luciferase cell culture lysis reagent (Promega). The lysates were centrifuged at 12,000 × g for 2 min at 4°C, and the supernatants were tested for luciferase expression using the luciferase reporter assay system (Promega) according to the manufacturer's instructions and a Victor3 multilabel counter (Perkin-Elmer).

siRNA experiments.

The following small interfering RNAs (siRNAs) were synthesized by Eurogentec: β-arrestin1 (5′-AAGCCUUCUGCGCGGAGAAU-3′) and β-arrestin2 (5′-GGCUUGUCCUUCCGCAAAGAC-3′) (28). HDMECs were transfected by using an Amaxa nucleofector kit for HUVECs according to the manufacturer's instructions. Epithelial cells were transfected by using a DharmaFECT kit (Dharmacon). An off-target siRNA (5′-GATAAGGTGCTGCGTGGACCC-3′) was used as a control. The cells were infected 3 days after transfection, and the efficiency of knockdown was assessed by immunoblotting with a rabbit monoclonal antibody against β-arrestin1 and β-arrestin2 (see the discussion of antibodies above).

Shear-stress experiments.

Cells were grown on disposable flow chambers (15μ-Slide VI [ibidi Gmbh], surface area = 0.6 cm² per channel) coated with 5 μg of rat tail collagen type I/cm². GFP-expressing N. meningitidis 2C43 was introduced and allowed to adhere for 10 min (10⁷ bacteria/slide). A continuous flow of prewarmed medium corresponding to a shear stress of 0.2 dyn/cm² was applied for 4 h using a syringe pump (Harvard Apparatus). This shear stress has been shown to impair bacterial adhesion on cells (15). Infected cells were then fixed with 4% paraformaldehyde for 20 min and washed twice prior to DNA and actin staining with DAPI and Alexa-conjugated phalloidin, respectively. For analysis of the 2D or 3D shape of the bacterial colonies, image acquisition was performed on an inverted microscope.

Immunoblotting.

Whole-cell lysate of cells were washed with ice-cold PBS and lysed in modified radioimmunoprecipitation assay buffer containing protease inhibitor cocktail (Sigma-Aldrich). Equal amounts of whole-cell lysates were then boiled and analyzed by SDS-PAGE. After transfer onto nitrocellulose (Optitran BA-S 83; Millipore), the blots were probed with the appropriate antibodies. Immunoblots were revealed by luminescence (ECL Plus; GE Healthcare Life Sciences).

RESULTS

Recruitment of junctional components by type IV pili is a common feature of N. meningitidis interaction with endothelial cells.

The crossing of the BBB by N. meningitidis has been studied using a cellular model of human cerebral microvascular endothelial cells (hCMEC/D3), which closely mimic the properties of the BBB such as the formation of tight junctions (39). As mentioned above, in this model, N. meningitidis recruits the polarity complex Par3/Par6/PKCζ and then the components of the cell-cell junctions underneath the microcolonies. This signaling leads to the formation of “gaps” between brain endothelial cells, by which the bacteria can transmigrate through the brain endothelium (4). In vivo, N. meningitidis colonies are also observed in close contact to peripheral endothelial cells, especially in the skin, where N. meningitidis induces typical purpuric lesions (7, 15). We reasoned that N. meningitidis colonies adhering on peripheral endothelial cells breach the vascular wall using the same mechanisms as the one observed on brain microvascular endothelial cells.

We investigated the delocalization of junctional proteins on primary HUVECs and primary HDMECs. The components of the adherens junctions (p120-catenin and VE-cadherin) and tight junctions (zonula occludens 1, claudin 5, and occludin) were monitored in HUVECs and HDMECs by immunostaining. These proteins were clearly visible at the cell-cell interfaces of uninfected cells, with the exception of occludin in HDMECs or claudin 5 in HUVECs. Cells were infected for 2 h with the capsulated and piliated N. meningitidis strain 2C43, which does not express Opa or Opc, the noncapsulated isogenic strain expressing Opa proteins (SiaD⁻ Opa⁺), or the nonpiliated noncapsulated isogenic strain (PilE⁻ SiaD⁻ Opa+) at an MOI of 100. The components of adherens junctions and tight junctions were recruited underneath piliated N. meningitidis microcolonies, where they colocalized with ezrin, i.e., the hallmark of the formation of the cortical plaque (Fig. 1A and Table 1). The recruitment of ezrin and that of the proteins of the adherens junction and tight junction is mediated by type IV pilus-dependent signaling (Fig. 1B).

Fig 1.

N. meningitidis recruits junction-like domains in peripheral endothelial cells. (A) Confluent monolayers of HDMECs and HUVECs were infected with the capsulated and piliated N. meningitidis strain 2C43 for 2 h. The components of tight junctions (ZO1) and adherens junctions (p120-catenin and VE-cadherin) and ezrin were immunostained as described in Material and Methods. Ezrin and both components of the adherens and tight junctions were recruited under the colonies. (B) HDMECs were infected with either a noncapsulated Opa-expressing strain (SiaD⁻ Opa⁺) or a nonpiliated noncapsulated, Opa-expressing strain (PilE⁻ SiaD⁻ Opa⁺). Only piliated strains recruit ezrin and the components of cell-cell junctions. (C) Cells were transfected with Par6-YFP 24 h before infection and then infected with N. meningitidis for 2 h. Par6-YFP was recruited under the microcolonies and colocalized with ezrin immunostaining. (D) HDMECs were transfected with a plasmid containing YFP alone and infected for 2 h. There were no recruitment of YFP protein under colonies. Scale bar, 10 μm.

Table 1.

Recruitment of ectopic junction-like domains in different cell lineages

Cell line	Recruitmenta
	Par3-myc	Par6-YFP	p120-catenin	E/VE-cadherin	ZO1	Claudin 3/5	Occludin
HCMEC/D3	+	+	+	+	+	+	+
HDMEC	ND	+	+	+	+	+	ND
HUVEC	+	+	+	+	+	ND	+
T84	−	−	−	−	−	−	−
Caco2	−	+/−	−	−	−	ND	−
Fadu	−	−
Detroit-562	−	−
Hec1b	−	−

^a+, >50% of the microcolonies recruited the protein; −, <10% of the colonies recruited the protein; +/−, <25% of the colonies recruited the protein; ND, no data available. Occludin was not immunostained in HDMECs, claudin was not immunostained in Caco2 cells and HUVECs, and the expression of Par3-myc plasmid was too low to be expressed in HDMECs.

We then monitored the localization of Par3 and Par6, two components of the polarity complex Par3/Par6/PKCζ after infection. Their recruitment at the site of bacterium-cell interaction is a prerequisite to the delocalization of junctional components in brain microvascular endothelial cells (4). HUVECs and HDMECs were transfected with either Par3-myc or Par6-YFP. Par6-YFP was recruited under the colonies and colocalized with the cortical plaque in both lineages (Fig. 1B and Table 1). Par3-myc colocalized with ezrin under more than 50% of the microcolonies of infected HUVECs (Table 1). However, expression of transfected Par3-myc in HDMECs was not efficient enough to conclude that Par3-myc is recruited. To further study the opening of the paracellular route, we tested the formation of “gaps” between infected cells. HDMECs were grown to confluence and infected with the piliated and capsulated N. meningitidis strain 2C43 at an MOI of 100 for 5 h. Actin and ZO1 staining were used to monitor the cell-cell junction integrity during infection. As expected, the formation of microcolonies on endothelial cells was associated with the formation of gaps between infected cells (Fig. 2A). Thus, the ability to open the paracellular route by recruiting the polarity complex and the components of cell-cell junctions under microcolonies is a common feature to peripheral and cerebral endothelial cells.

Fig 2.

N. meningitidis specifically open endothelial cell-cell junctions. HDMEC (A) and T84 (B) monolayers were infected with N. meningitidis (strain 2C43). Actin was stained with phalloidin-488 (green) and ZO-1 was immunostained (shown in red). After 5 h of infection, gaps and junction defects were visible under or close to large N. meningitidis colonies in the HDMECs (arrows). On the other hand, ZO1 and actin structures in the T84 cells were not modified compared to those of the control monolayers. The same results were obtained with cells seeded on transwells (data not shown). The colonization of cells by bacteria is similar for both cell types (N. meningitidis colonized 18.9% ± 10.8% of the total HDMEC surface and 32.4% ± 11.0% of the total T84 cell surface). Scale bar, 10 μm.

The formation of cortical plaque in meningococcal infected epithelial cells is not responsible for a delocalization of junctional components.

Meningococcal interaction on epithelial cells is responsible as for endothelial cells for the formation of cortical plaque with accumulation under the colony of ezrin, actin, adhesion molecules and membrane receptors that are clustered in a honeycomb-like structure. However, previous reports obtained with capsulated and piliated N. meningitidis suggested that tight junctions were not altered by meningococcal interaction with epithelial cells and that bacteria were likely to cross epithelial monolayer using the transcellular route (23, 31). We subsequently aimed at deciphering the discrepancies observed in meningococcal host cell interaction. We first confirmed that the capsulated and piliated N. meningitidis strain 2C43 does not alter the junctions of a tight-junction-forming monolayer of epithelial cells using T84 and Caco2 cells (Fig. 2B and data not shown). Furthermore, in both epithelial cells lines, none of the junctional components colocalized with ezrin under the microcolonies (Fig. 3 and Table 1). We next tested the recruitment of components of the polarity complex Par3-myc and Par6-YFP in both cell types. Par3-myc was not recruited under microcolonies in T84 and Caco2 cells. Par6-YFP was not recruited in infected T84 and was barely visible in <25% of the microcolonies in infected Caco2 cells (Fig. 3B and Table 1). Similar experiments were performed using three other epithelial cell lines—Hec1b, FaDu and Detroit-562. None of them showed a significant recruitment of Par3-myc or Par6-YFP underneath N. meningitidis microcolonies (Table 1). Altogether, these data demonstrate that N. meningitidis, when interacting with epithelial cells, does not recruit the polarity complex neither delocalize junctional proteins at the site of bacterial cell interaction.

Fig 3.

N. meningitidis does not recruit junction-like domains in epithelial cells. (A) Confluent monolayers of T84 epithelial cells were infected with N. meningitidis (strain 2C43) for 2 h. The components of tight (ZO1) and adherens junctions (p120-catenin and E-cadherin) and ezrin were immunostained as described in Materials and Methods. None of the components of the adherens and tight junctions were recruited under the colonies. (B) Cells were transfected with Par6-YFP 24 h before infection and then infected with N. meningitidis for 2 h. Par6-YFP was not recruited under the microcolonies. Scale bar, 10 μm.

N. meningitidis-induced signaling in epithelial cells is independent of the β₂-adrenergic receptor/β-arrestin pathway in contrast to endothelial cells.

The β₂-adrenergic receptor was shown to be responsible for the recruitment of β-arrestins under N. meningitidis colonies, following the interaction of type IV pilus components with the N-terminal part of the receptor in brain endothelial cells. We subsequently looked for a similar signaling pathway in primary human dermal microvessel endothelial cells using a previously published strategy (3). Briefly, the incubation of cells with a saturating concentration of a β₂-adrenergic receptor agonist, such as isoproterenol, is not only responsible for the transduction of a signal leading, in the case of isoproterenol, to an increase in intracellular concentration of cyclic AMP but also, via the activation of the β-arrestin, is responsible for the specific endocytosis of the receptor, thus depleting the receptor from the surfaces of the cells. In brain endothelial cells, such a depletion of the receptor from the cellular membrane prevents N. meningitidis from interacting with its receptor and inducing cellular signaling (3). We preincubated HDMECs for 2 h with 10 μM isoproterenol (saturating concentration, 10 μM = 10 × K_D). This concentration induced receptor internalization into recycling endosomes as shown in Fig. 4A. HDMECs were then infected for 2 h with N. meningitidis in the presence of the agonist. This induced a drastic decrease in ezrin recruitment in N. meningitidis colonies (Fig. 4B). To further investigate this signaling pathway, we assessed the presence of β-arrestins underneath N. meningitidis colonies. HDMECs were transfected with β-arrestin1-myc or β-arrestin2-myc and infected the day after with N. meningitidis for 2 h. β-Arrestins are clearly recruited at the adhesion site of colonies in almost all infected cells (Fig. 4C). Furthermore, the knockdown of β-arrestins by siRNA was associated with a decrease in ezrin-recruiting colonies in HDMECs (Fig. 4D and E). These results, similar to those obtained with the brain endothelial cell line, are consistent with the involvement of the β₂-adrenergic receptor/β-arrestin pathway during N. meningitidis infection of dermal microvessel cells.

Fig 4.

The β₂-adrenergic receptor/β-arrestin pathway is essential only for endothelial cells signaling. (A) HDMECs and Hec1b cells were transfected with β₂-adrenergic receptor-YFP plasmid and stimulated with 10 μM isoproterenol (a specific β₂-adrenergic agonist) for 10 min. As expected, in stimulated cells, the β₂-adrenergic receptor is internalized in vesicles, whereas in nontreated cells the receptor shows a diffuse distribution. Scale bar, 10 μm. (B) Cells were preincubated with 10 μM isoproterenol for 2 h before infection (in order to induce β₂-adrenergic receptor internalization) and then infected with N. meningitidis for 2 h in the presence of the agonist. The results are expressed as the percentages of microcolonies recruiting ezrin ± the standard errors of the mean (SEM). *, P < 0.001 (Student t test). Ezrin recruitment was not affected by the desensitization of the β₂-adrenergic receptor in epithelial cells. (C) HDMECs, T84 cells, and Hec1b cells were transfected with β-arrestin2-myc and infected with N. meningitidis for 2 h. β-Arrestin2-myc was not colocalized with ezrin under N. meningitidis colonies in >70% of epithelial cells but was recruited under almost all of the colonies in endothelial cells. Scale bar, 10 μm. (D) The upper panel shows the comparative expression of β-arrestins between the different cell lineages. The lower panel shows the efficiency of siRNA knockdown against β-arrestin1+2. The cells were lysed 4 days after transfection with siRNA against β-arrestin1+2 (Si) or a scramble siRNA (Scbl) and were studied for β-arrestin expression by Western blotting with a specific β-arrestin antibody. The results are representative of three independent experiments. (E) Ezrin recruitment under N. meningitidis colonies after β-arrestin knockdown. The results are expressed as the percentage of microcolonies recruiting ezrin ± the SEM. *, P < 0.001 (Student t test). (F) Hec1b cells were cotransfected with the β₂-adrenergic receptor tagged with YFP and β-arrestin2-myc and infected with N. meningitidis for 2 h. In the same cell, β₂-adrenergic receptor was recruited under the colony, whereas β-arrestin2-myc was not recruited. Scale bar, 10 μm.

The activation of a similar pathway following N. meningitidis adhesion on epithelial cells was then tested. We first confirmed the expression and the functionality of the β₂-adrenergic receptor pathway at the plasma membrane of epithelial cells. As mentioned above, the consequence of the activation of the β₂-adrenergic receptor is an increase in the production of cyclic AMP (cAMP). Cells were transfected with the CRE-luciferase plasmid in which the luciferase gene is under the control of several cAMP response elements (CREs), thus providing a sensitive measurement of cAMP produced downstream of the β₂-adrenergic receptor activation. After 2 h of stimulation with a saturating concentration of isoproterenol, an agonist of the β₂-adrenergic receptor inducing the production of cAMP, both T84 and Hec1b cell lines showed a significant increase in luciferase production compared to that of nonstimulated cells, indicating that these cells express a functional β₂-adrenergic receptor at the plasma membrane (data not shown). Moreover, we showed that the YFP-tagged receptor is internalized after activation with isoproterenol (Fig. 4A and data not shown), indicating that the β₂-adrenergic receptor/β-arrestin pathway is functional in these cell lines. We also confirmed that each cell line expresses the β-arrestins (Fig. 4D). We then tested whether the receptor internalization inhibits the formation of the cortical plaques on T84 and Hec1b cells. The cells were incubated with isoproterenol for 2 h, infected for another 2 h, and assessed for ezrin recruitment by immunostaining. Interestingly, when T84 and Hec1b cells were infected by N. meningitidis, the incubation with isoproterenol did not inhibit the formation of the cortical plaque (Fig. 4B). To confirm these results, the recruitment of tagged β-arrestins in infected T84 and Hec1b cells was assessed. These cell lines were transfected with β-arrestin1-myc or β-arrestin2-myc plasmid and infected the following day with N. meningitidis for 2 h. β-Arrestins were rarely recruited underneath bacterial microcolonies (Fig. 4C). Finally, cells were transfected with siRNA against β-arrestins. The knockdown of the β-arrestins did not inhibit the N. meningitidis-induced signaling on Hec1b and T84 cell lines (Fig. 4D and E).

Considering that each cell line can express different level of the β₂-adrenergic receptor and can differentially modulate the accumulation of the receptor at the plasma membrane, we coexpressed the β₂-adrenergic receptor tagged with YFP and β-arrestin2-myc in Hec1b cells. We reasoned that a low receptor concentration at the plasma membrane may be responsible for a lack of activation of the β₂-adrenergic receptor/β-arrestin pathway. Interestingly, the YFP-tagged β₂-adrenergic receptor expressed in Hec1b cells is sequestrated under colonies, whereas β-arrestin2-myc staining is diffuse in the cytoplasm (Fig. 4F). This result indicates that N. meningitidis, even though able to recruit the β₂-adrenergic receptor, is unable to activate the β-arrestin pathway in the Hec1b cell line. Altogether, these data demonstrate that the β₂-adrenergic receptor/β-arrestin pathway, even though present, is not required for the formation of the cortical plaque in epithelial cell lines.

Actin polymerization underneath N. meningitidis microcolonies is Src independent in epithelial cells.

The tyrosine kinase Src (Src) is essential for actin polymerization in honeycomb structures under colonies on endothelial cells. The Src kinase is recruited and activated by the β₂-adrenergic receptor/β-arrestin pathway and the ErbB2 receptor (3, 8). Considering the results above demonstrating that the β₂-adrenergic receptor/β-arrestin pathway is not involved in the signaling in epithelial cells, we examined whether Src was involved in the formation of actin protrusions in epithelial cells and peripheral endothelial cells. HDMECs, T84 cells, and Hec1b cells were preincubated with either PP1 or PP2 (two different Src inhibitors) for 1 h before infection by the capsulated and piliated N. meningitidis strain 2C43. As expected, Src inhibition was associated with a decrease in actin honeycomb structure underneath microcolonies in HDMECs, but there was no effect of both Src inhibitors in infected epithelial cells (Fig. 5). This result confirmed that the “epithelial cell” cortical plaque is strikingly different from that of endothelial cells.

Fig 5.

Actin polymerization underneath microcolonies is Src independent in epithelial cells. Cells were preincubated with a 50 or 20 μM concentration of two different Src inhibitors (PP1 or PP2, respectively) for 1 h before infection with N. meningitidis. Actin was stained using Alexa-conjugated phalloidin. The results are expressed as the percentages of microcolonies associated with actin honeycomb ± the SEM. *, P < 0.001 (Student t test).

The epithelial cell response to N. meningitidis adhesion is not sufficient to protect the microcolonies from shear stress.

Considering that actin polymerization does not involve the Src pathway in epithelial cells, we investigated whether mechanical properties of the N. meningitidis-induced signaling in epithelial cells are different from those of endothelial cells. The actin polymerization at the site of bacterial cell interaction leads to the formation of microvillus-like structures. In endothelial cells, they are known to protect the colonies on the apical surface of cells from the shear stresses and are likely to play an essential role during the formation of these colonies in the blood flow (15, 19). We subsequently assessed whether the formation of these microvilli in epithelial cells was able to protect the colonies from shear stresses.

HCMEC/D3, HDMEC, and T84 cells were grown on ibidi chambers as described in Material and Methods and subjected to a laminar flow under an inverted microscope. Capsulated and piliated GFP-expressing N. meningitidis strain 2C43 was introduced into the chamber and allowed to adhere for 20 min without laminar flow; the cells were then submitted to a regular laminar flow creating a shear force of 0.2 dyn/cm². After 4 h of infection, the N. meningitidis microcolonies developed on T84 cells were flat, i.e., only bacteria directly in contact with T84 cells are forming the colony (2D shape), while on endothelial cells the N. meningitidis microcolonies were mainly in a 3D shape (Fig. 6). This suggests that on epithelial cells, unlike endothelial cells, most of the bacteria in the colonies are washed out by the flow shear forces and that only the epithelial cell-adhering bacteria are able to resist the flow.

Fig 6.

The epithelial cell response to N. meningitidis adhesion is not sufficient to protect the microcolonies from shear stress. HCMEC/D3 cells, HDMECs, and T84 cells were grown on ibidi chambers until confluence. The GFP-expressing N. meningitidis strain 2C43 was allowed to adhere for 10 min. Cells were then exposed to a continuous flow, creating a shear stress of 0.2 dyn/cm² for 4 h. (A) Actin and DNA were stained using Alexa-conjugated phalloidin and DAPI, respectively. Immunofluorescence was analyzed by using a confocal microscope, and a 3D reconstruction was performed. (Left panel) x-y axis; (right panel) 3D reconstruction of 40 (D3), 48 (HDMEC), and 76 (T84) z-stacks of 0.25 μm. Arrows show the z axis of the 3D reconstruction. Scale bar, 10 μm. Microcolonies grown on endothelial cells under flow have several layers of bacteria unlike epithelial cells where a single layer of bacteria adhering to the cells is observed. (B) After infection, microcolonies were observed throughout the ibidi chambers using an inverted fluorescence microscope and were classified according to their 2D shape or 3D shape. The results are expressed as the percentages of colonies in a 2D or 3D shape after three independent experiments. Differences in the proportion of 3D colonies or 2D colonies between both endothelial cell lineages and the T84 cell lineage were significant (P < 0.001, chi-square test) in both comparisons.

The minor pilin PilV is not the bacterial determinant involved in the signaling to epithelial cells.

The minor pilin PilV was described to be responsible for N. meningitidis-induced signaling in endothelial cells. In a study by Mikaty et al., PilV mutation was associated with a drastic decrease in ezrin recruitment under microcolonies, and a reduced resistance of cell-adherent colonies to shear stress (19). Moreover, PilV was shown to be able to recruit the signaling receptor (i.e., the β₂-adrenergic receptor) in a reconstituted model of infection using the HEK293 cell line overexpressing the β₂-adrenergic receptor (3). Because this signaling pathway is only induced in endothelial cells, we assessed the role of PilV in the N. meningitidis-induced signaling of epithelial cells. We infected T84 and Hec1b cells with the capsulated and piliated strain 2C43 and with a pilV mutant derivative. Ezrin recruitment was assessed by immunostaining and used as the hallmark of the cell signaling induced by N. meningitidis. The ability of the pilV mutant to induce cell signaling was not different from that of the wild-type 2C43 strain (Fig. 7), ruling out the role of the minor pilin PilV in the signaling of capsulated and piliated N. meningitidis to epithelial cells.

Fig 7.

The minor pilin PilV is not the bacterial determinant involved in the signaling to epithelial cells. Epithelial cells were infected with either the wild-type strain 2C43 or the isogenic pilV mutant derivative. The results are expressed as the percentages of microcolonies recruiting ezrin ± the SEM. T84, P = 0.07; Hec1b, P = 0.1 (Student t test).

DISCUSSION

The data reported in this study extend the initial observation made using a brain endothelial cell line to other primary peripheral endothelial cell types such as HUVECs and HDMECs, showing that upon N. meningitidis adhesion to endothelial cells there is activation of the β₂-adrenergic receptor/β-arrestin signaling pathway, leading to the recruitment below the colonies of the Src tyrosine kinase and junctional components and the formation of gaps between cells. On the other hand, N. meningitidis adhesion on epithelial cells did not activate the β₂-adrenergic receptor/β-arrestin signaling pathway, suggesting that the mechanisms by which N. meningitidis colonize epithelial cells are different from those of endothelial cells. Thus, the β₂-adrenergic receptor/β-arrestin pathway is either of no significance or redundant in the epithelial cell-N. meningitidis interplay. Consequently, Src tyrosine kinase, which is involved in the actin polymerization following the β₂-adrenergic receptor/β-arrestin pathway, is not required for the formation of the microvilli in epithelial cells. Consistently, the minor pilin PilV that recruits the β₂-adrenergic receptor is not essential to induce an epithelial response.

We examined the possibility that the level of β₂-adrenergic receptor expression is crucial for N. meningitidis-induced signaling. Interestingly, tagged β₂-adrenergic receptor is sequestrated under the bacterial colonies that adhere to Hec1b cells. However, the sequestration of this receptor does not imply a direct interaction with the bacteria and is likely to be indirect. In Hec1b cells, the tagged β₂-adrenergic receptor is not able to promote the recruitment and the activation of the β-arrestins. It highlights the role of the cellular context, such as posttransductional modification, coreceptor interaction, and homo- or heterodimerization.

It has been previously reported that type IV pili are responsible for bacterial adhesion and signaling and that the β₂-adrenergic receptor is only involved in the transmission of signal but not that of adhesion, suggesting that a different receptor is responsible for the physical adhesion of N. meningitidis on cells. This is consistent with previous data reporting the role of CD46 or I-domain-containing integrins as an adhesion receptor for type IV pili (5, 9). However, the role of CD46 in meningococcal adhesion has been challenged in several reports (9–11, 31, 32). Considering that bacterial adhesion occurs via a different receptor than the signaling receptor, this adhesion receptor may be considered as a coreceptor of the β₂-adrenergic receptor. N. meningitidis could interact with a different coreceptor on endothelial and epithelial cells. Such interaction between a coreceptor and a G-protein-coupled receptor (GPCR) has already been reported. For instance, gp120 of HIV first interacts with the CD4 protein of T helper lymphocytes. The gp120-CD4 interaction enables a second interaction with the GPCR CXCR4 or CCR5. Both CD4 and GPCR interactions are required to allow HIV entry into the target cell. It should be pointed out that these GPCRs and CD4 are known to directly interact at the plasma membrane independently of gp120 (40). The interaction of N. meningitidis type IV pili with a different coreceptor that the β₂-adrenergic receptor may have in epithelial cells will promote different consequences on the bacterial cell interaction.

It has been shown that the Opa proteins may act in synergism with the type IV pilus-mediated adhesion to ensure bacterial invasion in epithelial cells (26). A large number of Opa receptors, i.e., the CEACAM family receptors, is critical for Opa-CEACAM interaction in a capsulated background. We have previously showed that Opa mediated adhesion is sufficient to allow interaction of the type IV pilus with the β₂-adrenergic receptor in a reconstituted model of signaling (3). Although we have extensively used a strain that do not express Opa, further work will be needed to address the role of Opa-type IV pilus synergism during signaling to the host cell.

Another hypothesis to explain this difference may rely on pilus retraction. Indeed, several studies have described the role of pilus retraction in Neisseria gonorrhoeae-induced signaling on epithelial cells such as actin polymerization (13, 17). On the other hand, a meningococcal strain that is unable to retract pilus was described to signal and adhere normally to HUVECs under flow shear stresses (19). In addition, a mutant unable to retract pili recruited the cell-cell junction components when adhering on brain endothelial cells (4). This suggests that the consequence of pilus retraction may be different on endothelial and epithelial cells.

N. meningitidis is a commensal of the human nasopharynx. Consistently, the frequency of invasive infection remains very low compared to the high proportion of carriers. On the other hand, N. meningitidis is a highly efficient pathogen that crosses the BBB at a high frequency, since up to 50% of people with an invasive infection develop meningitis (25). Our in vitro data demonstrate that both events, crossing the nasopharynx and crossing the BBB, are likely to involve different cell signaling pathways leading to two different paths of interaction in two different niches.

(i) Bacterial aggregates per se, formed by bacterium-bacterium interactions, are unable to resist to flow shear stresses. However, when grown on endothelial cells, bacterial aggregates are able to resist very high levels of shear stresses, since the aggregates are protected by actin-enriched cellular projections that spread around individual bacteria (15). The data reported here suggest that the actin-rich cell protrusions formed in epithelial cells are unable to protect the microcolonies from shear stresses. We speculate that the epithelial plasma membrane projections may not extend deeply between bacterial cells in the microcolonies. In the nasopharynx, N. meningitidis microcolonies must be exposed to irregular but slow flow shear stresses and sometimes to high levels of shear stresses due to coughing. The opposite is observed in the bloodstream, where the shear stress level varies between 1 and 100 dyn/cm² and occasionally decreases to values below 0.5 dyn/cm² in capillaries (15). In the case of nasopharyngeal colonization, the colony does not need to be stabilized. Moreover, this can facilitate the spread of the bacteria in vivo to a new niche in the same host or between individuals. In the same manner, the meningococcus has developed specific tools to achieve this goal, such as the pptB-dependent phosphorylation of the pili that limits the bacterial aggregation and favors dissemination (2).

(ii) We clearly show here that the opening of the paracellular route is restrained to endothelial cells in the case of brain microvascular or peripheral vascular endothelial cells. On the other hand, there is no visible opening of tight-junction-forming epithelial monolayers of T84 or Caco2 cells. Our data are consistent with previously published works showing that the crossing of a tight-junction-forming epithelial monolayer occur via a transcytosis pathway and is not associated with a decrease in transepithelial resistance or a disruption of intercellular junctions (18, 23, 31). This is consistent with the low frequency of infection, although the opening of the endothelial wall seems to be very efficient and can participate in the peripheral symptoms observed in meningococcal infections.

After adhesion, N. meningitidis can be found in the intracellular epithelial compartment (23, 24, 30). Several studies suggested that internalized capsulated bacteria are able to survive and even multiply within cells (12, 14, 21, 29). Sutherland et al. have shown that crossing a polarized epithelial monolayer depends on an intact microtubule network, suggesting that there is an intracellular trafficking across epithelial cells (31). It should be pointed out that the related pathogen Neisseria gonorrhoeae has been shown to migrate through the epithelial cells (38). These data are consistent with our results and do not support a paracellular crossing of an intact epithelium, although they confirmed the role of the β₂-adrenergic receptor/β-arrestin pathway in endothelial cells.