Report

Meningococcal Type IV Pili Recruit the Polarity Complex to Cross the Brain Endothelium

See allHide authors and affiliations

Science  03 Jul 2009:
Vol. 325, Issue 5936, pp. 83-87
DOI: 10.1126/science.1173196

Abstract

Type IV pili mediate the initial interaction of many bacterial pathogens with their host cells. In Neisseria meningitidis, the causative agent of cerebrospinal meningitis, type IV pili–mediated adhesion to brain endothelial cells is required for bacteria to cross the blood-brain barrier. Here, type IV pili–mediated adhesion of N. meningitidis to human brain endothelial cells was found to recruit the Par3/Par6/PKCζ polarity complex that plays a pivotal role in the establishment of eukaryotic cell polarity and the formation of intercellular junctions. This recruitment leads to the formation of ectopic intercellular junctional domains at the site of bacteria–host cell interaction and a subsequent depletion of junctional proteins at the cell-cell interface with opening of the intercellular junctions of the brain-endothelial interface.

Neisseria meningitidis is a commensal bacterium of the human nasopharynx that, after bloodstream invasion, crosses the blood-brain barrier (BBB) (1). Few pathogens have a tropism for the brain, indicating that N. meningitidis possess specific components to interact with the BBB. Meningeal colonization by invasive capsulated N. meningitidis is the consequence of the bacterial adhesion onto brain endothelial cells (2, 3), which is followed by bacterial division onto the apical surface of the cells (movie S1). This process is mediated by type IV pili (Tfp) (49). In addition, by powering a form of cell locomotion reported as twitching motility (10), Tfp lead to the spread of the bacteria on the surface of the cells and the formation of microcolonies. Subsequent to the formation of these microcolonies, Tfp trigger the recruitment of cortical actin and signal transducing proteins, leading to the formation of filopodia-like structures (2, 1113). The crossing of the BBB by N. meningitidis implies that, after Tfp-mediated adhesion, the bacteria transcytose through the brain capillaries and/or open the brain endothelium.

To investigate whether adhesion of N. meningitidis affects the integrity of adherens junctions (AJs) and/or tight junctions (TJs) of human brain endothelial cells, we analyzed the consequences of infection by N. meningitidis on the distribution of junctional proteins by using the human brain microvascular endothelial cell line hCMEC/D3 (14). After infection, components of the AJ (VE-cadherin, p120-catenin, and β-catenin) and the TJ (ZO1, ZO2, and claudin-5) were targeted underneath N. meningitidis colonies (Fig. 1A). At the site of N. meningitidis adhesion, these junctional proteins codistributed with each other and with the actin honeycomb-like network. In noninfected cells, the recruitment of junctional proteins usually occurs at the cell-cell interface and is controlled by several polarity proteins (Par3/Par6/PKCζ) (1517). In infected monolayers, Par3 and Par6 were observed underneath N. meningitidis colonies (Fig. 1B). Thus, N. meningitidis triggers a signal leading to the formation of an ectopic domain containing filopodia-like structures and enriched in junctional proteins, thus resembling spotlike AJs observed during early steps of junctional biogenesis. We refer to this domain as an “ectopic early junction-like domain” (18). By using isogenic derivatives, Tfp-induced signaling was shown to be responsible for the formation of these ectopic early junctionlike domains (fig. S1, A and B). However, Tfp retraction through the PilT motor was not required for formation of the ectopic domains (fig. S1, D and E).

Fig. 1

N. meningitidis recruits ectopic junctionlike domains beneath colonies. (A) VE-cadherin (green), the main component of the endothelial AJ, colocalized with actin (red) beneath N. meningitidis (Nm) colonies (top row). Two other AJ components, p120-catenin and β-catenin, and three components of the TJ, ZO-1, ZO-2, and claudin-5, are recruited under N. meningitidis colonies (bottom row). Arrow indicates a bacterial colony. Scale bars indicate 10 μm. (B) Yellow fluorescent protein (YFP)–tagged Par6 (par6-YFP) or myc-tagged Par3 (par3-myc), both green, are recruited underneath N. meningitidis colonies where they colocalize with actin (red). Areas outlined in white indicate the presence of a N. meningitidis colony. Scale bars, 10 μm. The formation of these ectopic early junctionlike domains is not found underneath all N. meningitidis colonies. Signaling underneath bacterial microcolonies required a minimal number of 20 bacteria per colony to be detected by immunofluorescence, with around 40 to 50% of microcolonies containing 40 to 50 bacteria. The average number of colonies signaling after 2 hours of infection is 40%.

The small guanosine triphosphatase Cdc-42 is required for polarization of mammalian cells (19, 20). The role of this component in the recruitment of the polarity complex by N. meningitidis was investigated. Transfection of a dominant negative mutant of Cdc42 or knockdown of Cdc42 by RNA interference (RNAi) inhibited the recruitment of Par6, Par3 (Fig. 2A and fig. S2A), VE-cadherin, p120-catenin, and actin (Fig. 2B and figs. S2B and S3). These results link the Cdc42/polarity complex pathway with the formation of the ectopic early junctionlike domains.

Fig. 2

The Cdc42-Par3/Par6/PKCζ pathway controls the formation of ectopic early junctionlike domains. Data are expressed as mean ± SEM. (A) Knockdown of Cdc42 was performed by using specific small interfering RNA duplexes (Cdc42 siRNA). Cells were cotransfected with par6-YFP or par3-myc. Knockdown of Cdc42 by RNAi reduced the recruitment of par6-YFP and par3-myc by fourfold. *t test (P < 0.005). (B) Knockdown of Cdc42, Par6 and Par3 were performed as described (27) (Cdc42 siRNA, Par6 siRNA, and Par3 siRNA). Scramble siRNA and siCONTROL were used as control for Cdc42/Par6 and Par3 knockdown, respectively. Knockdown of Cdc42 by RNAi reduced the recruitment of VE-cadherin, p120-catenin, and actin by 2.2-fold, 2.3-fold, and 2.5-fold, respectively. See also fig. S3. Knockdown of Par6 by RNAi reduced the recruitment of VE-cadherin, p120-catenin, and actin by 2.7-fold, 2.4-fold, and 2.4-fold, respectively. Knockdown of Par3 by RNAi reduced the recruitment of VE-cadherin by twofold. *t test (P < 0.01), **t test (P < 0.002). (C and D) HCMEC/D3 cells were either incubated with 3 μM or 6 μM of PKCζ-PS or PKCη-PS (control) or left untreated. (C). PKCζ-PS (6 μM) reduced VE-cadherin, p120-catenin, and actin recruitment by 8.5-fold, 5-fold, and 4.9-fold, respectively. *t test (P < 0.001). (D) HCMEC/D3 cells were transfected with either par6-YFP or par3-myc. Par3-myc recruitment was reduced 9-fold by 6 μM PKCζ-PS, but par6-YFP recruitment was not affected [*t test (P < 0.001), **t test (P < 0.01)].

The role of the polarity complex in the recruitment of junctional proteins was further explored by studying the inhibition of Par3 and Par6 with either dominant negative mutants or knockdown by RNAi. PKCζ inhibition was assessed by using a PKCζ pseudosubstrate inhibitor (PKCζ-PS) (21). Inhibition of Par6 and PKCζ reduced the recruitment of p120-catenin, VE-cadherin, and actin (Fig. 2, B and C, and figs. S2C and S3) and that of Par3 (Fig. 2D and fig. S2E), consistent with the finding that the Par6/PKCζ complex recruits Par3 at intercellular junction domains (22). On the other hand, inhibition of Par3 reduced only the recruitment of VE-cadherin (Fig. 2B and figs. S2D and S3), consistent with Par3 being specifically needed for junctional proteins targeting at early cell-cell junctions (23). These observations confirmed the role of the polarity complex in the recruitment of the junctional proteins by N. meningitidis.

The sequence of events leading to the targeting of AJ proteins at the cell-cell junctions during cellular polarization remains unknown. To get insight into this process, we engineered a VE-cadherin knockdown of hCMEC/D3 cells by stable expression of a VE-cadherin short hairpin RNA (VEC shRNA) (Fig. 3, A and B, and fig. S4A). In this cell line, p120-catenin and actin were still recruited beneath N. meningitidis colonies, whereas recruitment of β-catenin was dramatically reduced. On the other hand, down-regulation of p120-catenin using RNAi (Fig. 3C, S4B) resulted in inhibition of VE-cadherin and of actin recruitment. Consistent with a previous report, cortactin and Arp2/3 were not recruited by the bacterial colonies in p120-catenin knockdown cells (24) (fig. S4C). Furthermore, inhibition of Src kinase, which phosphorylate cortactin and is activated after the formation of the cortical plaque (25), did not modify p120-catenin recruitment but inhibited VE-cadherin and actin recruitment (fig. S4, D and E). Taken together, these results strongly suggest that p120-catenin–mediated recruitment of actin and VE-cadherin requires the recruitment and phosphorylation of cortactin by the Src kinase. In summary, Cdc42, via the polarity complex, organizes this ectopic early junctionlike domain, mainly by the initial recruitment of p120-catenin.

Fig. 3

P120-catenin is key to the recruitment of both actin and AJ proteins. (A and B) VE-cadherin silencing was performed by stable expression of a VE-cadherin shRNA (VEC shRNA). (A) Recruitment of β-catenin, p120-catenin, and actin was determined by immunofluorescence. Knockdown of VE-cadherin had no effect on the recruitment of p120-catenin and actin but reduced β-catenin recruitment by 20-fold. *t test (P < 0.001). Data are expressed as mean ± SEM. (B) In VEC-shRNA–expressing cells, p120-catenin was still recruited beneath N. meningitidis colonies, where it colocalized with actin (top), whereas β-catenin was no longer recruited (bottom). Areas outlined in white indicated the location of a N. meningitidis colony. Scale bars, 10 μm. (C) Silencing of p120-catenin was performed by using a specific siRNA duplex (p120 siRNA). Recruitment of VE-cadherin and actin was determined by immunofluorescence. Knockdown of p120-catenin reduced VE-cadherin and actin recruitment by 10-fold and 4-fold, respectively. *t test (P < 0.001). Data are expressed as mean ± SEM.

We asked whether the signal triggered by Tfp and leading to the formation of these ectopic early junctionlike domains destabilized intercellular junctions, especially by redirecting a recycling pool of junctional proteins to the N. meningitidis adhesion site. First, inhibition of protein synthesis did not prevent recruitment of VE-cadherin (fig. S5A). Second, inhibition of clathrin-coated pit formation blocked VE-cadherin recruitment (fig. S5, B and C), suggesting that VE-cadherin internalization is required for its targeting underneath N. meningitidis colonies. Third, when monolayers were tagged before infection with a VE-cadherin monoclonal antibody, antibodies were relocalized beneath colonies in infected monolayers (fig. S6). Thus, the VE-cadherin delocalized by the bacteria was coming from the intercellular junctions. This redistribution of the AJ proteins was associated with a reduction of the amount of tagged VE-cadherin at the intercellular junction (fig. S6 and movie S2). Thus, the junctional VE-cadherin is internalized and then mistargeted at the site of bacterial cell interactions.

Depletion of intercellular junction proteins from the cell-cell interface could open a paracellular route for bacterial spread. Indeed, N. meningitidis was shown to increase permeability to Lucifer Yellow (LY), a compound that marks passive paracellular diffusion (Fig. 4A) (26). Moreover, this increase relied on PKCζ activity and bacterial piliation (Fig. 4A). This modification of permeability was associated with the formation of gaps between infected cells (Fig. 4B). The number of gaps increased over time and was reduced by the PKCζ-PS (Fig. 4, B and C). Gaps did not form when cells were infected with a nonpiliated strain, showing that these gaps are due to Tfp-mediated signaling (Fig. 4C). Indeed, piliated strain cross the monolayer at a rate higher than those of nonpiliated isogenic derivatives or a piliated strain in the presence of PKCζ-PS (Fig. 4D). Thus, the signaling induced by N. meningitidis Tfp that leads to the recruitment of the polarity complex is associated with large alterations of the intercellular junctions that are sufficient for the bacteria to cross the brain endothelial cell monolayer.

Fig. 4

N. meningitidis–induced PKCζ activity facilitates cell-cell junction opening. (A) The permeability coefficient of Lucifer Yellow was measured 4 hours postinfection by N. meningitidis (Nm) or its nonpiliated isogenic strain (Nm ΔpilE), or after treatment by PKCζ-PS or PKCη-PS (6 μM). N. meningitidis induced a 1.55-fold increase compare with control. D-mannitol, which disrupts all cell-cell junctions, induced a 3.1-fold increase. *t test (P < 0.001). (B) HCMEC/D3 cells were incubated with 6 μM of PKCζ-PS or of PKCη-PS (control). (a) VE-cadherin localization was analyzed on the basolateral cross section of N. meningitidis–infected cells. Yellow arrowheads and areas outlined in yellow indicate gaps between cells. Areas outlined in red indicate the presence of N. meningitidis colonies. Blue bars marked 1 to 4 refer to Z-axis reconstruction images 1 to 4 in (b). Scale bars, 20 μm. (b) Z-axis reconstructions from a stack of 0.12-μm interval images show that VE-cadherin is apically relocalized underneath N. meningitidis colonies (white arrows) only in cells treated with PKCη-PS (control). (C) HCMEC/D3 cells grown on 3.0μm pore size inserts were treated or not with PKCζ-PS and incubated with N. meningitidis or its nonpiliated isogenic strain (N.meningitidis ΔpilE). Size and quantity of gaps observed 4 hours after infection are calculated as described (27). (D) Diffusion of N. meningitidis through an hCMEC/D3 monolayer, 4 hours postinfection, is 3.2-fold higher than diffusion of N. meningitidis in the presence of 6 μM PKCζ-PS and 16.5-fold higher than diffusion of its nonpiliated derivative (Nm ΔpilE). The rate of N. meningitidis internalization, determined by gentamicin protection assay, is very low (1 colony-forming unit in 3.5 × 105), identical to that of a control without cells, thus excluding a possible transcytosis of bacteria. Data are expressed as fold increase of N. meningitidis diffusion and calculated as described (27). Data from (B) to (D) are one representative experiment of three independent duplicates. Data are expressed as mean ± SEM.

In summary, N. meningitidis microcolonies trigger via type IV pili a signal resembling the one responsible for the formation of AJ at cell-cell junctions. This leads to the formation of ectopic early junctionlike domains (fig. S7), thus disorganizing the cell-cell junctions and opening the paracellular route, which allows N. meningitidis to cross the BBB and to invade the meninges.

Supporting Online Material

www.sciencemag.org/cgi/content/full/1173196/DC1

Materials and Methods

Figs. S1 to S7

References

Movies S1 and S2

  • Present address: INSERM U970, Paris Cardiovascular Research Center, Paris F-75015, France.

References and Notes

  1. Materials and methods are available as supporting material on Science Online.
  2. The authors thank M. Drab, P. Martin, I. Allemand, and N. Simpson for reviewing the manuscript and M. Garfa-Traore and N. Goudin for technical support. M.C. was funded by la Fondation pour la Recherche Médicale (FRM).
View Abstract

Stay Connected to Science

Navigate This Article