Pathogenic Neisseria Hitchhike on the Uropod of Human Neutrophils

Polymorphonuclear neutrophils (PMNs) are important components of the human innate immune system and are rapidly recruited at the site of bacterial infection. Despite the effective phagocytic activity of PMNs, Neisseria gonorrhoeae infections are characterized by high survival within PMNs. We reveal a novel type IV pilus-mediated adherence of pathogenic Neisseria to the uropod (the rear) of polarized PMNs. The direct pilus-uropod interaction was visualized by scanning electron microscopy and total internal reflection fluorescence (TIRF) microscopy. We showed that N. meningitidis adhesion to the PMN uropod depended on both pilus-associated proteins PilC1 and PilC2, while N. gonorrhoeae adhesion did not. Bacterial adhesion elicited accumulation of the complement regulator CD46, but not I-domain-containing integrins, beneath the adherent bacterial microcolony. Electrographs and live-cell imaging of PMNs suggested that bacterial adherence to the uropod is followed by internalization into PMNs via the uropod. We also present data showing that pathogenic Neisseria can hitchhike on PMNs to hide from their phagocytic activity as well as to facilitate the spread of the pathogen through the epithelial cell layer.


Introduction
The genus Neisseria includes the two obligate human pathogens N. gonorrhoeae and N. meningitidis. N. gonorrhoeae primarily colonizes the urogenital mucosa, where it crosses the intact mucosal barrier and remains in the tissue to initiate an inflammatory response [1]. N. meningitidis is commonly found in the nasopharynx of healthy individuals, where it can cross the mucosal epithelium and cause sepsis and/or meningitis. Bacterial dissemination across the blood brain barrier (BBB) has been reported to occur by hijacking the badrenoreceptor/b-arrestin pathway, triggering the opening of the intercellular junctions of the brain-endothelial interface [2].
Bacteria can also enter the meninges from the nasopharynx by redistribution of the intracellular junction protein N-cadherin in olfactory epithelia [3]. Although these two species of bacteria cause different diseases, the molecular mechanisms during infection are similar and they share many important virulence factors involved in both adhesion to and invasion of epithelial cells. Initial adherence by Neisseria to host epithelial cells is mediated by type IV pili (T4P) [4,5], which elicit cortical plaque formation in epithelial cells [6]. It has been reported that T4P of pathogenic Neisseria bind two different types of receptors: the complement regulator CD46 [7,8] and the I-domain-containing integrins on epithelial cells [9]. However, the role of CD46 in adhesion of Neisseria is yet not fully elucidated [10]. Upon infection, CD46 is phosphorylated at tyrosine 354, clusters beneath microcolonies of N. gonorrhoeae and induces a transient release of Ca 2+ from intracellular stores [11,12,13]. Bacterial adhesion to epithelial cells depends on the major pilus subunit protein PilE [14,15] and the pilus associated protein PilC [16,17,18,19]. PilC undergoes intrastrain phase variation [17] and has been suggested to be the tipadhesin of the pilus fiber [20]. In N. gonorrhoeae, both PilC1 (pilC2D)-and PilC2 (pilC1D)-expressing bacteria adhere to epithelial cells, while only PilC1 (pilC2D)-expressing N. meningitidis mediate adherence [21,22,23]. In contrast, Morand et al. reported that meningococcal PilC2 mediated adhesion to epithelial cells [19]. Initial adherence is followed by tight adherence where the Opacity proteins (Opa) in the bacterial outer membrane function as secondary bacterial ligands to host cell epithelium. Opa interacts with carcinoembryonic antigen-related cell adhesion molecules (CEACAM) and heparan sulfate proteoglycans (HSPG) on epithelial cells [24,25,26]. Infection by pathogenic Neisseria triggers release of pro-inflammatory chemokines that recruit polymorphonuclear leukocytes (PMNs or neutrophils) to the site of infection [27,28,29]. Although PMNs can engulf and kill most microorganisms, N. gonorrhoeae is remarkably resistant to PMN killing [26,30]. In fact, gonorrhea is characterized by a purulent urethral or cervical discharge consisting primarily of PMNs containing intact intracellular bacteria. In vitro assays show that PMNs only eradicate up to 70% of piliated N. gonorrhoeae added within the first hour [31]. N. gonorrhoeae has developed a number of strategies to evade the antimicrobial activity of PMNs, such as IgA protease-mediated degradation of lysosomal LAMP1 [32] and inhibition of signaling pathways required for inducing oxidative burst [26].
Severe meningococcal disease is characterized by early PMN activation. An accumulation of PMNs in the subarachnoidal space is one of the hallmarks of N. meningitidis infection [33]. Early neutropenia is associated with a high risk of developing severe meningococcal disease [34,35]. PMN activation enhances microbial clearance but also contributes to the vascular damage and multiorgan failure associated with meningococcal sepsis. N. meningitidis causes a moderate increase in CD11b (Integrin a M b2/complement receptor 3 a subunit) and CD18 (Integrin b2/ CR3 b subunit) in PMNs [36]. CR3 participates in pilus-mediated adherence to cervical epithelia [37].
Circulating PMNs isolated from blood are round and nonpolarized. PMNs become polarized upon chemotactic activation and contact with a matrix such as epithelial cells or coated glass.
The key element that drives cell motility and cytoplasmic remodeling is the activity of the cytomatrix [38]. This physiology of a polarized PMN can be divided into pseudopod, cell body and uropod. The pseudopod at the front of a migrating PMN mediates the phagocytosis of microorganisms. Hence, the uropod is the rear of a moving PMN and possesses no phagocytic activity. Uropods are enriched in adhesion-and signaling-receptors such as integrins, HPSGs, inter-cellular adhesion molecules (ICAMs), actin and ezrin/radixin/moesin (ERMs) [39]. PMNs also express CD46, but at 10-fold lower concentrations compared to epithelial cells [40].
Over 30 years ago, Densen and Mandell observed that single piliated gonococci adhered to the surface of PMNs while non-piliated bacteria were more easily engulfed [41]. Here, we reveal new data showing that pathogenic Neisseria adheres by its T4P to the uropod of human PMNs. Meningococcal adherence depends on both PilC1 and PilC2 for adhesion, while N. gonorrhoeae adherence does not. Using live cell imaging techniques, we demonstrate a new mechanism where pathogenic Neisseria can hitchhike on the uropod of PMNs and thereby avoid the phagocytic activity of the PMNs and simultaneously increase their ability to spread at the epithelial cell layer.

N. gonorrhoeae and N. meningitidis adhere to human PMNs
We analyzed the ability of Neisseria to adhere to human PMNs. Freshly isolated PMNs were first allowed to adhere to the glass bottom of cell culture dishes. The uropod was distinguished from the pseudopod by the fact that the uropod is at the rear of the migrating PMN. Bacteria were then added to the polarized PMNs at a multiplicity of infection (m.o.i.) of 100. Within 30 minutes of incubation, both N. gonorrhoeae MS11 and N. meningitidis FAM20 adhered to the uropod of viable PMNs. We followed the interaction by live-cell microscopy for 3 hours during which images were captured every 5 seconds. Bacterial adherence was reversible and whole bacterial microcolonies or parts of a microcolony could disassociate from the uropod. Gonococcal and meningococcal adherence to the uropod occurred either by direct uropod contact or by transport on the plasma membrane from the pseudopod to the uropod. Figure 1A shows a representative image sequence of a microcolony first encountering the pseudopod but evading phagocytosis by migrating on the plasma membrane to the uropod (Fig. 1A, Movie S1). Furthermore, this adherence to the plasma membrane was pilus-mediated, as determined by scanning electron microscopy (SEM) ( Fig. 1B and C).
Microcolonies of different sizes and single bacterium could adhere to the uropod of polarized PMNs but microcolonies consisting of approximately ten to one hundred bacteria were more frequently observed. The adherences of a single meningococcus, a small microcolony and a big microcolony are shown in Figure 2A. Identical adherence patterns were seen for gonococci. We confirmed the bacterial-uropod interaction by SEM and transmission electron microscopy (TEM) ( Fig. 2B and C, respectively). In Figure 2C, uropod adherence is captured at a later stage when there is a tighter interaction between the bacteria and the PMN prior to internalization (close up). Other bacteria were also captured prior to phagocytosis by the pseudopod protrusions.
Adherence of MS11 and FAM20 to the uropod was also observed in differentiated HL60 cells (Fig. S1), a promyeoloblast cell line that differentiates into neutrophil-like cells after chemical treatment and exhibits phagocytic activity and responsiveness to chemotactic stimuli (Methods S1).
Taken together, these data show that piliated translucent colonies of N. gonorrhoeae as well as piliated Opa-positive N. meningitidis adhere to the uropod of human PMNs. Bacteria that encounter other sites of the PMN plasma membrane also use their pili to promote adhesion.

Bacterial adherence occurs frequently and does not impair PMN velocity
Next, we quantified the frequency of uropod-specific bacterial adherence. The interactions between PMNs and bacteria were monitored during 1 to 3 hours of incubation by live-cell imaging and at least 100 cells were monitored. At each time point, images of randomly selected microscopic fields were captured and the PMNs were counted. No differences were found between PMNs carrying single bacterium or microcolonies when quantifying the frequency of uropod adherence. However, PMNs carrying the amount seen in Figure 2A (middle) was the most common. Already after 1 hour, N. gonorrhoeae MS11 bound more frequently to the uropod than N. meningitidis FAM20. The number of PMNs with adhering bacteria on the uropod increased during the incubation and after 3 hours, 16% of the PMNs observed had bound N. meningitidis FAM20 and 33% had bound N. gonorrhoeae MS11 (Fig. 3A). The experiment was repeated three times with PMNs isolated from different blood donors. The frequency of adherence varied slightly between the blood donors. The number of bacteria adhering to the uropod varied with the size of the microcolony (as shown in Fig 2A) and was dynamic all through the analysis. Hence, the number of adhering bacteria per PMN was not calculated.
We investigated whether the burden of a big bacterial microcolony on the uropod impaired PMN motility. The velocity of a PMN carrying a big microcolony on the uropod was measured in live-cell conditions using a tracking module. The motility of the PMN was not affected since both PMNs alone and PMNs with adhering bacteria moved on glass with a velocity of about 0.15 to 0.20 mm/s (Fig. 3B).
The pilus promotes bacterial adherence to the uropod and depends on PilC for binding Since pilus interactions with host cells play a major role in bacterial adherence to epithelial cells, we also determined the role of T4P in uropod adherence to human PMNs. To visualize T4P in living cells, piliated bacteria were labeled with a fluorescent DyLight 488 NHS ester and imaged by Total Internal Reflection Fluorescence (TIRF). With this technique, the total reflection of a laser generates an evanescent field that yields high resolution within a few nanometers above the object glass. Hence, the fluorescently labeled pili become visible [42]. Single bundles of fluorescent T4P could be detected at the uropod in TIRF within 60 minutes of bacterial incubation (Fig 4A, Movie S2). This was never observed in non-piliated gonococcal or meningococcal strains (data not shown). These data clearly showed that pili extending from the microcolonies directly mediated binding between the bacteria and the uropod of viable PMNs. The pilus binding to the uropod was confirmed in SEM electrographs (Fig. 4B). Thus, T4P mediate binding between the bacteria and the uropod of a PMN.
Taken together, these data show that T4P mediate bacterial adherence to the uropod of human PMNs while non-piliated Neisseria interacts very infrequently with PMNs. Meningococcal uropod adherence depends on both PilC1 and PilC2, while Uropod adherence is specific for Neisseria species Non-pathogenic commensal Neisseria species were also tested for PMN adherence. Neisseria subflava and Neisseria lactamica both exhibited adherence comparable with that of MS11 P-n, i.e., to less than 5% of the PMNs (Table 1). However, occasional pili-like structures were detected on these two strains by TIRF (Fig. S2, Methods S1) but none of these non-pathogenic strains express PilC [17]. Moreover, Streptococcus pyogenes, Escherichia coli and the type IV pili-expressing Pseudomonas aeruginosa did not adhere to the PMNs (Table 1).
Next, we analyzed the importance of Opa proteins in bacterial uropod adherence. Several morphologically different Opa-positive colonies were selected from N. gonorrhoeae MS11 under a binocular microscope. Opa-positive bacteria were incubated with PMNs as described previously. No alterations in adherence were seen for different MS11 P + , Opa + bacteria compared to MS11 P + , Opabacteria (data not shown). However, these results do not include some Opa-expressing bacteria that do not increase colony opacity. We therefore examined the adherence of translucent MS11 P-n colonies, which would include both Opa-negative and translucent Opa-positive bacteria. Although extensively repeated, no difference in adherence was observed. The role of these bacteria will be of future interest. We also tested whether bacteria or PMNs have to be viable for bacterial adherence. Indeed, no uropod-specific adherence could be detected after 3 hours of incubation when PMNs or bacteria were fixed with paraformaldehyde prior to infection (data not shown).

CD46 accumulates at the uropod upon bacterial adherence
We hypothesized that a cellular component important for uropod-specific adherence might be enriched at the uropod. Consequently, we first analyzed the role of CD46 as a pilus receptor of PMNs. CD46 was blocked with polyclonal antibodies against CD46 before infection but no inhibition of bacterial adherence was detected (data not shown). However, immunofluorescence staining with antibodies revealed that CD46 had accumulated at the uropod adjacent to the adhering bacteria (Fig. 5). Adherence of FAM20 P + and the PilC2-expressing mutant FAM20 DpilC1 induced a redistribution of CD46 to the uropod, while that of FAM20 DpilC2 did not (Fig. 5). Hence, the recruitment of CD46 at the uropod depends on PilC2. Uninfected PMNs and PMNs incubated with the non-piliated MS11 Pn showed equally distributed CD46 at the cell surface (Fig. 5).
I-domain-containing integrins function as pili receptors important for gonococcal adherence to epithelial cells [9]. Therefore, we also investigated the role of both CD11b (integrin a1) and CD29 (integrin b1) in bacterial-uropod interaction. Pre-incubation with antibodies against either CD11b or CD29 did not reduce bacterial adherence (data not shown). Also, there was no accumulation of CD11b or CD29 at the uropod ( Fig. 6A and B).
Taken together, these data show that CD46 is recruited beneath the site of meningococcal adhesion and is PilC2-dependent.  During gonococcal adherence, both PilC1 and PilC2 expression is required for successful CD46 redistribution.

Intracellular bacteria at the uropod indicate bacterial internalization
While imaging bacterial adherence to the PMNs by SEM, electrographs frequently showed intracellular diplococci at the uropod, clearly visible underneath the thin layer of the uropod plasma membrane (Fig. 7A). These findings led us to further investigate the possibilities of bacteria entering the PMNs. TEM electrographs showed bacteria in membrane enclosed compartments both at the pseudopod and at the uropod (Fig. 7B). However, the bacteria at the uropod appeared to be in a visually discernible compartment, compared to the phagosomal compartments at the pseudopod (Fig. 7B, close up). To further investigate this in living cells, DyLight 488 NHS-stained bacteria were incubated with freshly isolated PMNs stained with Lysotracker only as contrast (and not for identifying lysosomes). Bacteria were taken up by the uropod (Fig 7C, Movie S3) although phagocytosis by polarized cells is primarily attributed to the pseudopod.
Therefore, these experiments suggest that bacteria can enter the PMN in two different ways: (1) by engulfment/phagocytosis or (2) by bacteria-promoted internalization.
PMNs transport bound bacteria across epithelial cell layers N. meningitidis adheres to the human pharyngeal epithelium. Pharyngeal cells release cytokines that recruit PMNs to the site of infection. N. meningitidis FAM20 were allowed to form microcolonies and adhere to epithelial pharyngeal FaDu cells (ATCC HTB-43, LGC Standards). After 90 minutes of infection, freshly isolated PMNs were added to the cell culture medium. PMNs migrated towards microcolonies and although initial contact was made with the pseudopod, the microcolony ended up bound to the uropod. After this, the PMN was either immobilized and trapped by the microcolony that bound both to the epithelial cell and the uropod (Fig 8A, Movie S4) or the PMN was able to remove the microcolony from the epithelial cell and transport it away from the site (Fig 8B, Movie S5). The nature of PMNs allows them to easily migrate under and in between layers of epithelial cells. PMNs with microcolonies attached at the uropod were seen to migrate in between and under a tightly confluent layer of FaDu cells, potentially disrupting the junctions between them (Fig 8C, Movie  S6).
These data show how bacteria can use the uropod for transport and thereby spread over an epithelial cell surface with the help of PMNs in vitro, suggesting a mechanism that bacteria may utilize during an infection. This observation demonstrates a new route for Neisseria to penetrate epithelial layers.

Discussion
Recruitment of PMNs from the blood is an early event during bacterial infection. Upon bacterial adherence, host target cells release cytokines that promote PMN infiltration into infected tissue. Pathogenic Neisseria encounters high numbers of PMNs during colonization and close contact is inevitable between these cells. Bacterial pathogens often manipulate the immune system to their advantage [44,45] and N. gonorrhoeae is particularly specialized in evading killing by PMNs. Bacterial adherence to the uropod of actively migrating PMNs could be an additional mechanism for bacterial spread and immune evasion. A schematic illustration of the events is presented in Figure 9A.
In this study, we present a novel adherence mechanism for pathogenic Neisseria, namely pilus-mediated adherence to the uropod of PMNs. Freshly isolated PMNs from human blood were incubated with bacteria and the interaction was characterized and visualized by live-cell microscopy, TIRF, TEM and SEM. The direct pilus-PMN interaction was clearly shown in both live-cell conditions and high-resolution imaging. Previous studies have yielded contradictory results regarding the role of pili in adhesion and phagocytosis [46,47]. However, we observed that when a piliated bacterium encountered a pseudopod, it was transported on the plasma membrane to the uropod, thereby evading phagocytosis (Fig. 9A), a phenotype not seen in non-piliated strains. Although some bacteria evaded phagocytosis in this way, we also noticed that the same cell could phagocytose other piliated bacteria. It has previously been shown that protein-coated latex beads can be transported from the pseudopod to the uropod in a similar manner [48,49].
The adherence to PMNs is more frequent for gonococci than meningococci, corresponding to what is seen with epithelial cells. No adherence was detected for E. coli, S. pyogenes or type IVexpressing P. aeruginosa, suggesting that uropod adherence is specific for Neisseria. High levels of adherence were exclusive for pathogenic Neisseria, which implies that the uropod adherence is important during infection.
Piliated uropathogenic E. coli (UPEC) has been shown to bind uniformly to the surface of PMNs. The adhesion is indirectly mediated by type I pili that leads to CD55 receptor binding [50].
CD55 shows high structural similarity and functionality with CD46 and both glycoproteins are regulators of complement activation. We show that CD46 accumulated at the uropod in close proximity to the adhering Neisseria. CD46 does not have a direct role in bacterial adherence to the uropod. However, the role of CD46 in signaling should be investigated further. It is possible that the recruitment of CD46 may function in cellular signaling prior to uptake via the uropod or regulate the production of nitric oxide and interferon as seen in macrophages [51].
In N. meningitidis, expression of both PilC1 and PilC2 were required for adherence, indicating that PilC1 and PilC2 may have two different functions, either indirectly in the biogenesis of the pilus filament or directly binding to PMNs in vitro. Specific adhesion-promoting regions in PilC1 are most likely involved in this adherence as previously reported [52]. It is possible that PilC1 mediates the adherence as seen in epithelial cells, while PilC2 is involved in the CD46 signaling observed here. This is not due to variance in piliation since inactivation of either pilC1 or pilC2 results in the same number of pili [53]. PilC2 exhibits defined cell and tissue tropism [54] and correlates with the CD46-independent binding of PilC2-expressing T4P seen in epithelial cells [55]. Gonococcal adherence occurred in both PilC1-expressing and PilC2-expressing isogenic mutants, showing no differences between PilC1 and PilC2 for uropod adherence.
Intracellular bacteria were often present inside the uropod. Since pilus-mediated adherence of pathogenic Neisseria to epithelial cells is followed by invasion, we investigated the possibility that bacteria also invade PMNs. Although it is clearly shown here that bacteria can be internalized in some way via the uropod, it is difficult to separate the number of intracellular bacteria that have entered the phagosome via the pseudopod from those that have entered via the uropod since there is always an unknown time factor for the maturation of the phagolysosome. Bacterial entry via the uropod could be a strategy for pathogenic Neisseria to impair phagosomal maturation, thus leading to intracellular survival.
The observed interactions between bacteria and PMNs at the epithelial cell layer revealed the impact of bacterial adhesion to the uropod other than phagocytosis evasion. Addition of PMNs to epithelial cells and bacterial microcolonies impaired the PMNs' ability to migrate and phagocytose bacteria (Fig. 9B). The immobilization was transient and PMNs adhering to the microcolonies already bound to epithelial cells could break away from the epithelial cells and migrate with the attached bacteria away from the site of adhesion, therefore spreading bacteria to uninfected cells. It is possible that the removal of the bacteria by the PMN requires the cooperative action of several PMNs and that it depends on PMN load in the assay (Fig. 9C). Also, PMNs could disrupt cell-to-cell contact and transport bacteria in between the cells, hence potentially giving the bacteria access to the subepithelium (Fig. 9D). Several mechanisms have been described that explain how Neisseria cross the epithelium: through damaged epithelial layers; the transcellular route; depletion of junctional proteins followed by an opening at the cell-cell interface; and via a ''Trojan horse'' manner inside phagocytes [56,57,58,59]. Furthermore, the influx of leukocytes at the BBB [60] during meningococcal infections may be a possible way of hitchhiking bacteria to reach the CSF. It is generally assumed that pilus expression plays a major role in the crossing of the BBB by bacteria and is in agreement with data showing the importance of PilC-expressing bacteria for crossing the BBB [61].
In summary, pathogenic Neisseria hitchhike on the uropod and thereby utilizing PMNs for immune evasion and dissemination.

PMN isolation and cell lines
Non-coded blood from donors was purchased from the Blood Center and used within 2 hours after collection. Viable PMNs were isolated as described elsewhere using LympholyteH-poly (Biosite) [62]. The isolation method yielded pure PMNs that were .95% viable as measured by trypan blue staining. Freshly isolated PMNs were maintained in Dulbecco's modified eagle medium (DMEM, Invitrogen) supplemented with 20% inactivated fetal bovine serum (FBS, Gibco) and were always used within 4 hours after isolation. The epithelial pharyngeal cell line FaDu (ATCC HTB-43, LGC Standards) was maintained in DMEM supplemented with 10% inactivated FBS at 37uC and 5% CO 2 .

Bacterial strains
Strains used in this study are listed in table 2. Bacteria were grown on GCB (GC medium base Difco, Detroit, MI, USA) agar plates containing Kellogg's supplement [4] at 37uC and 5% CO 2 . Piliated (P + ), translucent phenotypes of gonococcal strains were distinguished by colony morphology under binocular light microscope and re-streaked every 18 to 20 hours to ensure a P + , Opaphenotype. N. meningitidis wild-type strain FAM20 belongs to serogroup C and is P + [63]. All bacterial strains determined to be P + have previously been analyzed to express T4P, as determined by electron microscopy and the ability to form microcolonies in solution (data not shown). After adherence assays, bacteria were harvested from the GC agar plate and re-suspended in DMEM. Optical density was measured at 600 nm to calculate the number of bacteria per milliliter.

Live cell imaging
Freshly isolated 2.5610 5 PMNs in DMEM (20% FBS) were added to glass bottom 12 well dishes (MatTek Corp.). The PMNs were incubated for 30 min at 37uC and 5% CO 2 , allowing the cells to adhere to the glass bottom. The cell culture dishes were transferred to a humidified incubation chamber (37uC, 5% CO 2 ) connected to an inverted fluorescence microscope (Cell observer, Carl Zeiss). Upon assay, 2.5610 7 bacteria were carefully added to each of the wells. The interaction was monitored in differential interference contrast (DIC) light for 0 to 3 hours. Images of randomly selected fields of vision were captured at different time points. Live-cell time-lapse movies were generated by capturing images of the cells every 5 seconds for 10 minutes. Images and movies were further processed by ImageJ software (National Institutes of Health) and Adobe Photoshop software (Adobe). No differences between bacterial adherence, PMN motility or phagocytic activity could be detected when cells were grown on poly-D-lysine-coated glass or serum-coated glass.

Transmission and scanning electron microscopy
For SEM, freshly isolated 2.5610 5 PMNs in DMEM (20% FBS) were first allowed to adhere to 13 mm uncoated cover glasses (nr. 0, Mentzel) and were then incubated with 2.5610 7 N. meningitidis FAM20 P + for 3 hours. Cells were fixed in 2.5% glutaraldehyde (Sigma) for 30 minutes and dehydrated by increasing concentrations of ethanol (30%, 50%, 70%, 90% and finally three times in 100% ethanol). The glasses were immersed in 100% hexamethyldisilazane (HMDS, Sigma) for 3 minutes and then dried in a moist free environment. The samples were coated with gold using a sputter coater and examined using a JEOL JSM-7000F.
For TEM, PMNs were incubated with N. gonorrhoeae MS11 P + (m.o.i. 100) for 3 hours at 37uC 5% CO 2 . The cells were washed three times with PBS (Accugene) and fixed in 3.7% paraformaldehyde (PFA, diluted in PBS) (Sigma). Cells were then rinsed in 0.1 M phosphate buffer (PB) and transferred to 2.5% glutaraldehyde (Sigma) in 0.1 M PB and stored in a refrigerator. Cells were then post-fixed in 2% osmium tetroxide in 0.1 M PB, pH 7.4, at 4uC for 2 hours, dehydrated in ethanol followed by acetone and embedded in LX-112 (Ladd, Burlington, Vermont, USA). Sections were contrasted with uranyl acetate followed by lead citrate and examined in a Tecnai 12 Spirit Bio TWIN transmission electron microscope (Fei, The Netherlands) at 100 kV. Digital images were taken by a Veleta camera (Olympus Soft Imaging Solutions, GmbH, Münster, Germany) [64].

TIRF microscopy
Freshly isolated 2.5610 5 PMNs in DMEM (20% FBS) were allowed to adhere to 35 mm poly-d-lysine-coated glass bottom dishes (MatTek corp.) for 30 minutes. One bacterial colony was resuspended in 30 ml of sterile PBS. To visualize bacterial proteins, whole viable bacteria were stained with the DyLight 488 NHS ester (Thermo Scientific). The DyLight staining did not affect any bacterial properties. DyLight 488 NHS was added to the bacterial suspension for 5 to 10 minutes at 37uC and 5% CO 2 (700 mg/ml in PBS). NHS-labeled bacteria were then added to the PMNs and observed under the microscope. Visualization of the pilus interaction with PMNs during live-cell time-lapse analysis was performed with total internal reflection fluorescence (TIRF) microscopy using a connected argon laser and a 1006 objective (N/A 1.46, Carl Zeiss).  the infected cells and PMNs was monitored by DIC light microscopy over time.

Statistical analysis
Student's t-test was used for PMN velocity analyses and data were considered significant if P,0.05. A one-way ANOVA was performed in bacterial adherence assays. Data were considered significant if P,0.05.

Supporting Information
Methods S1 Supporting material and methods. Movie S1 Bacteria escape engulfment by the PMN and move along the plasma membrane to the uropod. Movie shows the interaction between a bacterial microcolony of N. meningitidis FAM20 and a PMN. DIC images were captured at 0.2 Hz through a 63 x oil objective. The image sequence was further processed into a movie by ImageJ software. (AVI) Movie S2 Pili directly interact with the uropod. DyLight 488 NHS-stained N. meningitidis FAM20 were incubated with freshly isolated PMNs in glass bottom dishes. Interaction between the pili and PMNs was observed under the microscope in TIRF using a connected argon laser and a 100x oil objective (N/A 1.46). Fluorescence images were captured at 80 Hz. The image sequence was further processed into a movie by ImageJ software. (AVI) Movie S3 Bacteria are internalized at the uropod. DyLight 488 NHS-stained N. meningitidis FAM20 (green) were incubated with freshly isolated Lysotracker-stained (red) PMNs in glass bottom dishes. Cellular DNA was stained with Hoechst 33342 (blue). Bacterial adhesion at the uropod of the activated PMN is followed by internalization of the bacteria. Fluorescence images were captured at 0.1 Hz through a 100x oil objective. The image sequence was further processed into a movie by ImageJ software.

(AVI)
Movie S4 A PMN is trapped and immobilized by a bacterial microcolony. N. meningitidis FAM20 were allowed to form microcolonies and adhere to FaDu cells. After one hour, freshly isolated PMNs were added. Movie shows a bacterial microcolony bound to both an epithelial cell and a PMN. DIC images were captured at 0.2 Hz through a 63 x objective. The image sequence was further processed into a movie by ImageJ software. (AVI) Movie S5 PMN interacts with bacteria adhering to epithelial cells. N. meningitidis FAM20 were allowed to form microcolonies and adhere to FaDu cells. After one hour, freshly isolated PMNs were added. Movie shows a PMN removing a microcolony from the epithelial cell and transporting it away from the site of initial adherence to FaDu cells. DIC images were captured at 0.2 Hz through a 63 x objective. The image sequence was further processed into a movie by ImageJ software.

(AVI)
Movie S6 A PMN transports bacteria in between a cell monolayer. N. meningitidis FAM20 were allowed to adhere to FaDu cells. After one hour of infection, freshly isolated PMNs were added to the cell culture medium. Movie shows a microcolony adhering to a PMN. The PMN penetrates the layer of cells, transporting the bacteria between the cells. DIC images were captured at 0.2 Hz through a 63 x objective. The image sequence was further processed into a movie by ImageJ software. (AVI)