Disease-Associated Neisseria meningitidis Isolates Inhibit Wound Repair in Respiratory Epithelial Cells in a Type IV Pilus-Independent Manner

Abstract

Neisseria meningitidis is the causative agent of meningococcal disease. Onset of meningococcal disease can be extremely rapid and can kill within a matter of hours. However, although a much-feared pathogen, Neisseria meningitidis is frequently found in the nasopharyngeal mucosae of healthy carriers. The bacterial factors that distinguish disease- from carriage-associated meningococci are incompletely understood. Evidence suggesting that disruptions to the nasopharynx may increase the risk of acquiring meningococcal disease led us to evaluate the ability of disease- and carriage-associated meningococcal isolates to inhibit cell migration, using an in vitro assay for wound repair. We found that disease-associated isolates in our collection inhibited wound closure, while carriage-associated isolates were more variable, with many isolates not inhibiting wound repair at all. For isolates selected for further study, we found that actin morphology, such as presence of lamellipodia, correlated with cell migration. We demonstrated that multiple meningococcal virulence factors, including the type IV pili, are dispensable for inhibition of wound repair. Inhibition of wound repair was also shown to be an active process, i.e., requiring live bacteria undergoing active protein synthesis.

INTRODUCTION

Neisseria meningitidis, the causative agent of meningococcal disease, remains one of the top infectious killers worldwide (1). An estimated 500,000 individuals, many of them children, are affected each year (2). Healthy, asymptomatic individuals frequently carry meningococci in the nasopharyngeal mucosa; however, on occasion, the bacteria can breach the epithelial barrier and reach the bloodstream of susceptible individuals, causing bacteremia. If the bacteria cross the blood-brain barrier and reach the meninges, meningitis results. N. meningitidis is among the leading causes of meningitis in infants, adults, and the elderly, where ca. 10% of cases are fatal; among survivors, approximately 10% will suffer permanent hearing loss or other serious sequelae (3).

In spite of the serious diseases that N. meningitidis can cause, it is normally a benign component of the normal nasopharyngeal microflora for a significant proportion of the general population; carriage rates typically range from about 10 to 25%, with higher rates of carriage seen among young people in crowded conditions (4). Occasionally epidemics, lasting years or decades, can occur. From 1991 to 2008 New Zealand experienced one of these, with the majority (>85%) of disease cases during the epidemic period caused by a single strain type, defined as B:4:P1.7-2,4, belonging to the ST-41/44 clonal complex (5, 6). The prevalence of a single strain type made it feasible to generate a tailor-made serogroup B N. meningitidis vaccine (MeNZB), which was introduced to control the epidemic (7, 8). The epidemic situation, with a single strain type predominating, contrasts with the heterogeneity of carriage-associated isolates (9, 10). The bacterial factors that distinguish a highly virulent strain, such as the NZ epidemic strain type, from more benign isolates that are generally associated with carriage, remain incompletely characterized. Many N. meningitidis virulence factors aid in colonization and are equally present in disease- and carriage-associated isolates (11).

The mechanism by which N. meningitidis breaches the epithelial layer is not known. The olfactory nerve has been posited as a route of infection of the meninges (12), while the M cells of the nasal-epithelium-associated lymphoid tissue, in the adenoids and tonsils, may also provide a portal of entry (13). While intracellular invasion of epithelial cells by N. meningitidis has been observed and well characterized (14), the significance of this invasion in the development of carriage versus disease remains unclear (1, 15). There is epidemiological evidence that N. meningitidis may gain access to deeper tissues following disruptions or damage to the epithelial barrier of the throat (13). The strongest evidence for this is from the “meningitis belt” of Africa, where rates of epidemic meningococcal disease can be extremely high (16, 17). These epidemics have been shown to relate to the weather conditions (18). During the dry season, when the winds are very strong and levels of dust and other particulates in the air are very high, epidemics occur far more frequently; the incidence of disease has been shown to correlate precisely with the rate of wind speed, and the winds generally carry large amounts of irritants that can affect the throat (18). Additional irritants that increase the risk of acquisition of invasive disease include respiratory illnesses, which often precede the development of meningococcal disease (19). Similarly, cigarette smoke has been shown to raise the risk of meningococcal disease in both passive and active smokers (20).

We hypothesized that the ability to breach the epithelial barrier could be one trait that differs between disease- and carriage-associated N. meningitidis isolates. Multiple bacterial pathogens have been shown to inhibit wound repair in the host (21–24). We used a variation of a well-characterized in vitro wound repair assay (25, 26) to measure the ability of disease- and carriage-associated isolates to inhibit wound repair, demonstrating that inhibition of wound repair is carried out by disease-associated meningococci, is not dependent on type IV pili, and is a process that requires live bacteria undergoing active protein synthesis.

MATERIALS AND METHODS

Bacterial strains.

The majority of the disease-associated N. meningitidis isolates described here are part of the Meningococcal Reference Collection (MRC), collected and maintained by the Institute of Environmental Science and Research (ESR), as part of the surveillance of meningococcal disease in New Zealand on behalf of the Ministry of Health. The carriage isolates described in the present study (designated NZCM) were collected as part of a household contact carriage study, carried out in Auckland, New Zealand, in the late 1990s (27). The Neisseria lactamica isolate was also collected as part of this study. The N. gonorrhoeae isolate tested was the World Health Organization (WHO) C reference strain, from the Australian Gonococcal Surveillance Programme. Additional strains associated with the Burkina Faso 2002 epidemic and the 2000 Hajj epidemic were kindly provided by Dominique Caugant (28). N. meningitidis isolates are routinely typed using serological and sequencing methods to determine serogroup and serosubtype (porA allele). In addition, selected strains have been further analyzed, with the serotype (porB allele) determined by serology and sequence type (ST) by multilocus sequence typing, as described elsewhere (10, 29, 30). All isolates in the MRC are immediately frozen down at −80°C following minimal laboratory passage. Frozen working stocks are maintained of frequently accessed isolates to prevent repeated freeze-thaw cycles. N. meningitidis isolates are grown on commercially acquired Columbia blood agar (CBA) plates (Fort Richard Laboratories, Auckland, New Zealand) at 37°C in a humidified 5% CO₂ incubator (for routine passage) or on BBL brain heart infusion agar (Oxoid) plates, supplemented with kanamycin (50 μg/ml) where required. Escherichia coli strains DH5α (subcloning efficiency) or Top10 cells (both from Life Technologies) were used for DNA manipulation and were grown on Luria-Bertani agar plates supplemented with kanamycin (50 μg/ml) or ampicillin (100 μg/ml). Table 1 lists the isolates used in the present study.

TABLE 1.

Clinical and carriage isolates used in this study

Bacterial strain	Strain information (description, serogroup, serotype, serosubtype, sequence type)a	Source or reference
N. meningitidis
NZ98/254	Representative New Zealand epidemic strain type and target strain for MeNZB vaccine; B:4:P1.7-2,4; ST-42	50
NZCM107	Serogroup B carriage isolate; B:nt:19-15; ST-178	This study
N 28/00	Hajj epidemic isolate, from Norway in 2000; W:2a:5,2,36-2; ST-11	D. Caugant (28)
BuFa 15/02	Burkina Faso isolate from 2002 epidemic; W:2a:5,2,36-2; ST-11	D. Caugant (28)
NZ97/192	New Zealand epidemic strain type; B:4:P1.7-2,4; ST-154	This study
NZ96/329	Disease isolate from New Zealand; C:2b:P1,2; ST-66	This study
NZCM237	Carriage isolate from New Zealand; NG:nt:P1.16; unknown ST	This study
NZCM127	Carriage isolate from New Zealand. NG:4:nst; unknown ST	This study
NZCM158	Carriage isolate from New Zealand; NG:1:P1.22,14-6,36-2. ST-2154	This study
NCZM148	Carriage isolate from New Zealand; NG:nt:P1.16; ST-5159	This study
NZCM238	New Zealand epidemic strain type, isolated from a healthy carrier (household contact of patient); B:4:P1.7-2,4; ST-154	This study
NZCM97	Carriage isolate from New Zealand; B:15:P1.13; ST-198	This study
NZCM222	Carriage isolate from New Zealand; B:nt:P1.14; ST-162	This study
NZCM155	Carriage isolate from New Zealand; Y:14:P1.5,2; ST-23	This study
NZCM122	Carriage isolate from New Zealand; NG:15:P1.6; ST-198	This study
NZCM1	Carriage isolate from New Zealand; E:nt:P1.7,16; unknown ST	This study
NZCM235	Carriage isolate from New Zealand; A:nt:P1.5; unknown ST	This study
NZCM137	Carriage isolate from New Zealand; NG:14:1.6; ST-4183	This study
NZCM157	Carriage isolate from New Zealand; B:14:P1.14; unknown ST	This study
NZCM233	Carriage isolate from New Zealand; B:14:P1.7,1; ST-5102	This study
NZCM69	Carriage isolate from New Zealand; B:15:nst; unknown ST	This study
NZCM113	Carriage isolate from New Zealand; C:4:nst; ST-278	This study
N. gonorrhoeae	WHO C reference strain	Australian Gonococcal Surveillance Program
N. lactamica	Carriage isolate from New Zealand (NZCML001)	This study

^aNG, nongroupable; nt, nontypeable; nst, nonsubtypeable; ST, sequence type.

Cell culture and infection.

The bronchial respiratory epithelial cell line 16HBE14o− (abbreviated as 16HBE) was used for the majority of our studies of the interactions between N. meningitidis and host cells (31). 16HBE cells were routinely cultured in M199 medium supplemented with 10% inactivated fetal calf serum (FCS).

For the cell migration assays (described in further detail below), 16HBE cells were suspended at a density of 3 × 10⁵ cells/ml in M199 plus 10% FCS, with 100 μl of cell suspension added to each well of a collagen-coated Oris cell migration assay 96-well plate (Platypus Technologies, Madison, WI), which was incubated overnight to allow the cells to adhere. All wells in Oris cell migration plates contain a stopper insert to prevent the cells from adhering to a small section in the center of the well. For the in vitro scratch assay, 1-ml portions of 16HBE cells were seeded into 24-well plates at a density of 2.5 × 10⁵ cells per well, followed by incubation for 24 h to allow the bacteria to adhere to the plates.

Primary human tracheal epithelial cells (HTEpC) were obtained from PromoCell (Germany). Cells were isolated from normal tissues and were obtained from approved medical centers according to strict ethical standards. The cells were cultured according to the supplier's directions. Briefly, cells were cultured in airway epithelial cell basal medium, supplemented with 1× airway epithelial cell growth medium supplement pack (both obtained from PromoCell). The final concentrations of supplements were 0.004 ng/ml bovine pituitary extract, 10 ng/ml recombinant human epidermal growth factor, 5 μg/ml recombinant human insulin, 0.5 μg/ml hydrocortisone, 0.5 μg/ml epinephrine, 6.7 ng/ml triiodo-l-thyronine, 10 μg/ml human holotransferrin, and 0.1 ng/ml retinoic acid, under serum-free conditions. Primary tracheal epithelial cells were subcultured no more than four times and were cultured for a week prior to using for any assays. Oris collagen-coated cell migration assay 96-well plates were seeded with 100 μl of primary tracheal cells suspended at a density of 2 × 10⁵ cells/ml. The medium was changed daily.

For immunofluorescence microscopy, 16HBE cells were grown on 8-well glass slides (Nunc Lab-Tek II chamber slide system; Sigma) with each well seeded with 450 μl of cell suspension at 1.2 × 10⁵ cells/ml and grown until confluent (∼2 days).

For Rac and Rho activation assays, six-well tissue culture plates were seeded with 3 ml of 16HBE cell suspension at a density of 4 × 10⁴ cells/ml and grown for 3 days, reaching ca. 90% confluence.

Oris cell migration assay.

16HBE or primary tracheal epithelial cells were seeded onto collagen-coated Oris cell migration assay 96-well plates as described above, followed by incubation for the indicated times to allow the bacteria to adhere to the plates. After the incubation, the stoppers were carefully removed from all wells, apart from 8 to 10 wells, which retained the stoppers and were used to calculate the initial wound area. The medium was aspirated, the cells were gently washed twice with warm phosphate-buffered saline (PBS), and fresh medium (M199 medium with 10% inactivated FCS) was replaced. The cells were then returned to the incubator for about 2 h before the addition of bacteria. Overnight plate cultures of N. meningitidis were streaked to fresh plates in the morning and returned to the incubator for 2 to 3 h, after which the bacteria were scraped off and resuspended in M199 medium with 10% inactivated FCS. The optical density at 600 nm (OD₆₀₀) was read, and the suspension adjusted to an OD₆₀₀ of 0.002. Ten microliters of the bacterial suspension was added to each well, for an initial multiplicity of infection (MOI) of about 10. The plates were sealed with a Breathe-Easy sealing membrane (Sigma) to prevent interwell contamination and then incubated for 20 h at 36°C and 5% CO₂. After the incubation, the Breathe-Easy membrane was peeled off, and the remaining stoppers were removed. The supernatant from each well was aspirated and discarded. The cells were gently washed one time with 100 μl of warm PBS and fixed by adding 100 μl of ice-cold methanol to each well for 3 min. The methanol was aspirated, and the plates were air-dried for about 3 to 5 min. The wells were each imaged and photographed using an inverted microscope (Olympus) at the 4× setting. Images were analyzed using ImageJ (32) to calculate the remaining wound area as a percentage of the wells that had not had their stoppers removed.

In vitro wound repair scratch assay.

16HBE cells were seeded in 24-well plates as described above and incubated for about 24 h. After the cells had grown to ∼90% confluence, a single scratch was made down the middle of each well, using a sterile 20-μl pipette tip. The cells were washed three times with warm PBS to remove any cellular debris, and 1 ml of fresh medium (M199 plus 10% FCS) was added to the wells. The plates were returned to the incubator for 2 h. After 2 h, each well was photographed with an inverted microscope prior to the addition of bacteria; these images represent the initial wound area and were used to calculate the final extent of wound repair. Bacteria were cultured on plates overnight and streaked onto fresh plates in the morning, which were then incubated for 2 to 3 h. Bacteria from these plates were then suspended in M199 medium containing 10% inactivated FCS. The OD₆₀₀ was measured and adjusted to 0.002; 100 μl of the suspension was added to each plate, to yield an initial estimated MOI of about 10. The plates were covered with a Breathe-Easy sealing membrane (Sigma) and incubated for 20 h at 36°C at 5% CO₂. The following day the medium supernatants were aspirated and discarded. The cells were washed gently one time with 1 ml of warm PBS and fixed with 1 ml of ice-cold methanol for 3 min. The methanol was aspirated, and the plate was air-dried for 3 to 5 min. The wells were again photographed with an inverted microscope (Olympus), and the images were analyzed with ImageJ software (32); the degree of wound closure was calculated as a percentage of the initial wound area.

Actin staining of 16HBE epithelial cells.

16HBE cells were seeded onto eight-well glass slides as described above. Cultures that had reached confluence were serum starved for 24 h prior to wounding. Wounds were made with a 20-μl plastic pipette tip in a cross pattern; the cells were washed with PBS and placed in the incubator. After 2 h, wounded cultures were infected with N. meningitidis isolates similar to the method described when infecting wound repair assays, except that 50 μl of N. meningitidis culture at OD₆₀₀ of 0.002 were added to each well (MOI of 10). At 14 h postinfection, the cells were washed with cold PBS and fixed with 4% paraformaldehyde in PBS for 20 min at room temperature on the slide. Fixed cells were washed with PBS and permeabilized with 0.5% Triton X-100 in PBS for 10 min. Permeabilized cells were blocked with 10% heat-inactivated FCS in PBS for 30 min and stained with rhodamine-tagged phalloidin (20 μg/ml; Sigma) in blocking solution for 20 min. Slides were washed and mounted with Vectashield mounting medium containing DAPI (4′,6′-diamidino-2-phenylindole; Vector Laboratories) and stored in the dark at 4°C until analysis.

Cell morphology was examined on an Olympus upright fluorescence microscope (BX-51), and images were recorded with a digital camera attached to the microscope. To count the number of cells and lamellipodia along the wound edge, three to six pictures were taken at random positions along the wound edge for each well with a 20× objective lens. Images were merged and analyzed using ImageJ (32). The total number of cells along the wound edge were counted using the DAPI image, and the numbers of cells with lamellipodia were examined using the phalloidin image.

The correlation analysis between degree of wound healing and prevalence of lamellipodia was calculated using the average percentage of remaining wound, obtained from an in vitro scratch assay for each isolate (n ≥ 4). The Pearson correlation was calculated with the analysis function in GraphPad Prism5 software.

Enumeration of cell-associated bacteria.

The number of 16HBE cell-associated bacteria (i.e., adherent and intracellular) was quantified using a method based on the one described by Sutherland et al. (33). Briefly, 16HBE cells were cultured until confluent in six-well plates, at about 10⁶ cells/well. The normal medium was removed, the cells were washed with PBS, and serum-free medium was replaced, after which the cells were cultured another 16 to 20 h. The cells were infected with N. meningitidis at the same MOI as for the wound repair inhibition assays. The cells and bacteria were coincubated for 4 h, at which time the cells were washed three times with warm PBS to remove non-cell-associated bacteria. The cells were lysed with 500 μl of 1% saponin in PBS per well and incubated for 10 min with gentle shaking. The cell lysate was diluted and plated onto CBA plates, and the bacteria were enumerated. The input bacteria were also enumerated by plating a series of dilutions.

Rac and Rho activation assay.

The amount of Rac1- and RhoA-GTP was measured using a G-LISA Rac1 and RhoA Biochem kit (absorbance based; Cytoskeleton, Inc.) according to manufacturer's directions. Briefly, six-well plates were seeded with 1.25 × 10⁵ cells as described above and grown to 90% confluence. Cultures were serum starved for 24 h. Wounds were made with a 20-μl pipette tip in a 4×4 grid pattern to increase the number of wound edges in the culture. Cultures were infected with 400 μl of an N. meningitidis culture at an OD₆₀₀ of 0.002, as described above. At 5 min or 14 h after infection, the cells were washed with ice-cold PBS, and 400 μl of supplied lysis buffer with protease inhibitor (Complete protease inhibitor cocktail tablets, EDTA-free; Roche) was added to each well to extract protein. Protein extracts were frozen in liquid nitrogen and stored at −80°C. Protein concentration was estimated using the precision red advanced protein assay reagent included in the kit. Each assay used 50 μg of protein. Three biological replicates were analyzed for each isolate per assay. The levels of Rac1- or RhoA-GTP were presented as a normalized ratio of the value obtained for the positive control supplied by the manufacturer, which corresponded to 3 ng of Rac1-GTP or 1 ng of RhoA-GTP.

Construction of N. meningitidis mutants.

Mutants were made by allelic replacement of the targeted gene with an AphA3 kanamycin (Km) resistance cassette (34) by natural transformation. All primers were designed based on the published sequence of a serogroup B N. meningitidis strain, MC58 or NZ05/33, which is a representative isolate of the New Zealand epidemic strain type (35). Mutants were verified by PCR and sequencing. The siaD mutant, predicted to lack a capsule, was verified by performing a slide agglutination assay of the mutant and the wild type, using Neisseria meningitidis antisera for serogroups A, C, W, and Y (BD Difco). Antiserum to serogroup B, made against the NZ epidemic strain type capsule, was a generous gift from ESR. The ΔpilE ΔpilS mutant and wild-type parent were examined for type IV pili by transmission electron microscopy. Bacteria cultured on CBA plates were scraped into a 2% glutaraldehyde solution in PBS. The bacteria were then processed according to the protocol described by Burghardt and Droleskey (36) and imaged at the Otago Centre for Electron Microscopy.

Mutagenesis of porA, siaD, and pilC.

For each gene, both upstream and downstream flanking regions were amplified by PCR, using NZ98/254 genomic DNA as a template and the primers listed in Table 2. All internal primers include a long sequence of overlap with the Km primers. The AphA3 Km cassette in pUC19 (pUC19 Km) was used as a template for a third PCR, using the primers Km1 and Km2 (Table 2). The three PCR products were combined and used for a fusion PCR, amplifying with the outermost primers. The resulting fusion product was purified and cloned into the Topo TA vector (Life Technologies) according to the manufacturer's protocol. The resulting plasmid contained regions of homology for each gene, flanking the Km cassette; these plasmids do not replicate in N. meningitidis and were used for natural transformations of NZ98/254, as described below, to generate targeted mutants.

TABLE 2.

Primers used in this study

Primer	Sequence (5′-3′)	Feature(s)	Target gene or reference
Km1	GCGGAATTCGCCGTCTGAACCAGCGAACCATTTGAGG	EcoRI site	AphA3 Km cassette
Km2	GCGGAATTCGCTTTTTAGACATCTAAATCTAGG	EcoRI site	AphA3 Km cassette
PorA_1	GCGGGATCCCGGAAGGGTTTTGGTTTTT	BamHI site	Outer membrane porin PorA: NMB1429
PorA_2	GCGGGATCCACTTTTCGTCATTCCCACGA	BamHI site
PorA_3	GCTGGTTCAGACGGCGAATTCCGCGACAATACGAGGGCGGTAAG	Anneals to Km1; EcoRI site
PorA_4	AGATGTCTAAAAAGCGAATTCCGCGGTTTGCGCCACAAATTCTA	Anneals to Km2; EcoRI site
BCSiaD_1	GCGGGATCCCAGTAGCTTTAGGCGGTTCG	BamHI site	Alpha-2,8-polysialyltransferase SiaD; NMBNZ0533_0073
BSiaD_2R	GCGGGATCCGCCTTATAAACTTTACTACTGGGCTTT	BamHI site
BSiaD_3	GCTGGTTCAGACGGCGAATTCCGCTGTCAACCACATTGAATCTTGA	Anneals to Km1; EcoRI site
BSiaD_4	AGATGTCTAAAAAGCGAATTCCGCCAACTTGCCAGCAAAAACAA	Anneals to Km2; EcoRI site
PilC1_1R	GCGGGATCCGGACAAAGCCTGATCCTGCC	BamHI site	PilC1 homolog, NMBNZ0533_0475
PilC1_2R	GCGGGATCCTCAGTGAGATTGGCGAGATG	BamHI site
PilC_3_KmR	GCTGGTTCAGACGGCGAATTCCGCTTGCCGTAGGGCGGCAGGTAGG	Anneals to Km1; EcoRI site
PilC_4_KmR	AGATGTCTAAAAAGCGAATTCCGCCCCGAACGGATATGTTTACG	Anneals to Km2; EcoRI site
OpcA_F4	GCGGGATCCCACACCGACCTCTCTTCAT	BamHI site	Outer membrane protein OpcA; NMBNZ0533_1078
OpcA_R4	GCGCAATTGGAAGCGTATGCTTGGTGGTT	MfeI site
OpcA_B1M1-R1	GCGGGATCCCAATTGTTTCCAATCGGTTAAAATACGC	BamHI, MfeI sites
OpcA_F1BHI	GCGGGATCCAAAGAAGTTTTGTCGAGCAAGG	BamHI site
1971_F1BHI	GCGGGATCCCCGCCCTTCAAAATACACAT	BamHI site	Major outer membrane protein IB, PorB NMBNZ0533_1971
1971_R1KM1	GCTGGTTCAGACGGCGAATTCCGCCCTTTCAAACCGATGAAGGA	Anneals to Km1; EcoRI site
1971_F2KM2	AGATGTCTAAAAAGCGAATTCCGCCAATACGCGCTTAACGACAA	Anneals to Km2; EcoRI site
1971_R2BHI	GCGGGATCCAGTGCGTTTGGAGAAGTCGT	BamHI site
0338_F1BHI	GCGGGATCCGGCGCAATACAACGACATTA	BamHI site	Adhesion and penetration protein, hap (App), NMBNZ0533_0338
0338_R1KM1	GCTGGTTCAGACGGCGAATTCCGCGTCCGTTTGTCGGTTGTTTT	Anneals to Km1; EcoRI site
0338_F2KM2	AGATGTCTAAAAAGCGAATTCCGCTAAACGCCGAAATCAAAGGT	Anneals to Km2; EcoRI site
0338_R2BHI	GCGGGATTCGAGTGCACAAAGCCATCTGA	BamHI site
BHOmpAF1	GCGGGATCCGCGAAGTAAACAAGGCTTCG	BamHI site	OmpA family protein, NMBNZ0533_1864
OmpAR1	GCGGAATTCGTAGAAACATGGCGGCAAAT	EcoRI site
OmpAF2	GCGGGAATTCCGATGCATGAGGTTAGTGGA	EcoRI site
BHOmpAR2	GCGGGATCCATACAGTGCCGCACACATCT	BamHI site
lgtA-Bamh_F1_2	GCGGGATCCATTACGACTTATGCCCGCCT	BamHI site	Lacto-N-neotetraose biosynthesis glycosyltransferase NMBNZ0533_0398
lgtA-mfe_R1	GCGCAATTGCCCGTCAATAAATCTTGCGTA	MfeI site
lgtA-mfe_F2	GCGCAATTGTATGCAAAACCACGTTATCAGC	MfeI site
lgtA-hind_R2_2	GCGAAGCTTGTTCCGGCTGCAACATTACG	HindIII site
pilEF1-HindIII	GCGAAGCTTACAAGTTCTGATGTTCAGCTCGT	HindIII site	Fimbrial proteins, NMBNZ0533_0019 to NMBNZ0533_0031
pilER1-BamHI	GCGGGATCCACTTACCCCATTGATAAGGAA	BamHI site
pilEF2-BamHI	GCGGGATCCAACCATTTTACCGAACCGTCT	BamHI site
pilER2-SacI	GCGGAGCTCCACCGCCTTCAGTATAACCTA	SacI site
ivt-1	ATTGGCTCATAACACCCCTTGTATTA		37
ivt-2	GAACTTTTGCTGAGTTGAAGGATCA
Kan-2 FP-1	ACCTACAACAAAGCTCTCATCAACC		Provided with Epicentre EZ-Tn5 <Kan-2> kit
Kan-2 RP-1	GCAATGTAACATCAGAGATTTTGAG

Mutagenesis of opcA.

Regions of upstream and downstream homology were separately amplified by PCR, using NZ98/254 genomic DNA as a template and the primers listed in Table 2, which contained engineered MfeI restriction sites. The Km cassette was removed from pUC19 Km by digestion with EcoRI, which is compatible with MfeI. The PCR products were digested with MfeI, purified, combined with the EcoRI-digested Km cassette, and ligated together. The resulting ligated product was again digested with MfeI (to bias toward ligations that included the Km cassette, which have the MfeI site destroyed by religating with EcoRI ends). The ligation was amplified using the outermost primers. The purified and concentrated PCR product was used directly in a natural transformation of NZ98/254.

Mutagenesis of porB and hap (App).

For both genes, the upstream and downstream flanking regions were amplified from NZ98/254 genomic DNA, using the primers listed in Table 2. The Km cassette was also amplified from pUC19 Km using the Km1 and Km2 primers. The three resulting PCR products were purified and used as the template in a PCR fusion reaction, using the outermost primers. The resulting fusion PCR product was verified, further amplified and concentrated, and then used directly for a natural transformation of NZ98/254.

Mutagenesis of the NMBNZ0533_1864 (OmpA family protein), lgtA, and pilE pilS regions.

Both flanking regions were amplified from NZ98/254 genomic DNA, using the primers listed in Table 2. The Km cassette was also amplified from pUC19 Km, and all PCR products were purified. The upstream PCR product and the Km PCR product were combined and ligated at room temperature for 1 h; the resulting ligated product was amplified by PCR using primers against the upstream product and a Km primer and then purified. The resulting product was purified and combined with the purified 700-bp downstream product and then ligated as before. The resulting ligation product was finally amplified by PCR using the outermost primers. The final PCR product was further amplified, concentrated, and purified and then used for natural transformation of NZ98/254. For the pilE pilS region, a total of 8.1 kb were deleted from the genome, including genes NMBNZ0533_0019 through NMBNZ0533_0031. Although primer pilEF2-BamHI (Table 2) anneals to multiple loci in the genome, a band of ∼700 bp was gel extracted before proceeding with the ligations. Deletion of the pilE pilS region was verified using genomic PCR using the primers pilEF1-HindIII and pilER2-SacI (Table 2).

Mutagenesis of lgtC and pilM.

These two mutants were identified during a pilot experiment for generating an in vitro N. meningitidis transposon mutant library. Transposon mutants were constructed in the following way. Genomic DNA from NZ98/254 was purified and broken down into smaller fragments using the NEBNext dsDNA Fragmentase (NEB). A total of 2 μg of genomic DNA was combined with 10× buffer (provided with the kit) and 100× bovine serum albumin, with water added to bring the mixture to 36 μl. The reaction was vortexed thoroughly and incubated on ice for 5 min. Next, 4 μl of the Fragmentase enzyme was added to the reaction, which was vortexed again, and incubated at 37°C for 7 min and 30 s (to obtain DNA fragments in the 1- to 6-kb region). The reaction was stopped by the addition of 5 μl of 0.5 M EDTA. The digested DNA was purified with a Zymo DNA Clean & Concentrator column (Zymo Research) and eluted in a final volume of 10 μl. The efficacy of the reaction was checked by running digested or undigested genomic DNA on a gel. The digested targeted genomic DNA was subsequently used for an in vitro transposon reaction with an EZ-Tn5 <Kan-2> insertion kit (Epicentre Biosciences). Briefly, 1 μl of EZ-Tn5 10× reaction buffer was combined with 0.2 μg of digested genomic DNA, 1 μl of EZ-Tn5 <Kan-2> transposon, 1 μl of EZ-Tn5 transposase, and water to a final volume of 10 μl. The reaction mixture was incubated at 37°C for 2 h and stopped by adding 1 μl of EZ-Tn5 10× stop solution. The mixture was heated at 70°C for 10 min to inactivate the transposase. The mutagenized DNA was stored at −20°C. The transposon-treated DNA (0.2 μg) was used for a natural transformation of NZ98/254, according to our usual procedure (described below), with selection on BHI agar plates supplemented with kanamycin. Twelve colonies were chosen for further analysis. These were streaked onto fresh plates and frozen in glycerol broths. To determine the site of the transposon insertion, the RATE PCR was performed as described previously (37). Briefly, we combined 45 μl of Platinum PCR Supermix (Life Technologies) with 1 μl of primer ivt-1 or primer ivt-2 (Table 2) and 1 μl of suspended heat-killed N. meningitidis. After RATE PCR, the resulting products were purified on Zymo columns and sequenced with the Kan-2 RP1 or Kan-2 FP1 primers (included with EZ-Tn5 <Kan-2> kit; Table 2). One mutant was confirmed to have the transposon inserted in the lgtC gene (NMBNZ0533_0397) and one in the gene encoding the type IV pilus assembly protein PilM (NMBNZ0533_0519).

Natural transformations.

Natural transformations were performed using purified mutagenic plasmid or PCR products. Mutagenic plasmids and PCR products contained the homologous regions to the target gene, flanking the Km cassette. Natural transformations were performed essentially as described previously (38). Briefly, NZ98/254 N. meningitidis was grown overnight on CBA plates. The following morning, the overnight cultures were streaked onto fresh plates and returned to the incubator for 2 to 3 h. In the meantime, the mutagenic PCR or plasmid was prepared, usually by combining and concentrating 4 to 6 PCRs or plasmid preparations by using the Zymo DNA Clean & Concentrator-5 kit (Zymo Research) and suspending the concentrated DNA in about 20 μl of water or Tris-EDTA (yielding a typical concentration of about 350 ng/μl). A light loop of restreaked bacteria was then transferred to a small circular area on a fresh CBA plate, and the purified mutagenic DNA was added to the bacteria and mixed briefly with a sterile inoculating loop. The plates were incubated, facing up, for 4 to 5 h. After the incubation, the bacteria were swabbed with a sterile inoculating loop and suspended in 300 μl of M199 medium. The bacterial suspension was then divided and spread onto two plates of BHI agar supplemented with kanamycin. Any colonies resulting from overnight growth were streaked onto fresh plates before being frozen in glycerol broths. The correct insertion of the Km cassette was confirmed by PCR and sequence analysis of the targeted gene region.

RESULTS

Disease-associated N. meningitidis inhibit wound repair in respiratory epithelial cells in an in vitro assay.

We hypothesized that, if N. meningitidis was able to access the bloodstream via damaged host epithelial tissue, more virulent isolates would display an enhanced ability to inhibit wound repair compared to carriage-associated isolates. We selected a panel of low-passage-number disease-associated clinical isolates and carriage-associated isolates that were taken from healthy volunteers (see Table 1). All N. meningitidis strains were typed according to standard laboratory protocols, including sequencing and/or serology. Isolates recovered from healthy volunteers and not belonging to hypervirulent clonal complexes were considered carriage associated, although we cannot rule out that these strain types could be responsible for sporadic cases of disease. Rarely, isolates belonging to the New Zealand epidemic strain type were recovered from healthy individuals. The isolates considered disease associated were all recovered from disease cases in New Zealand or were representative isolates of overseas epidemics, including isolates from the Burkina Faso epidemic of 2002 and the Hajj epidemic of 2000 (28).

The ability of N. meningitidis to inhibit wound repair was measured by adding the bacteria to cells undergoing wound repair, using a well-characterized in vitro scratch assay (25, 26) or an Oris cell migration assay (Platypus Technologies). The tight-junction-forming human bronchial epithelial cell line 16HBE14o− (16HBE), which retains the differentiated morphology and function of normal human airway epithelia (31), was used in these assays. In the scratch assay, confluent cell monolayers were scraped with a 20-μl pipette tip. For the Oris cell migration assay, a physical block is placed in the center of each cell culture well; the cells are seeded and grown to confluence, after which the block is removed, leaving an empty area for cells to migrate into. In each case, bacteria were added after scratching the cells or removal of the physical block, and the progress of the migrating cells was assessed. Closure of the wound area within the first 24 h has been shown to be primarily due to cell migration, rather than proliferation (26).

Wounded cell cultures were infected with different N. meningitidis isolates and incubated; the remaining wound area was measured 20 h after infection; a no-bacteria control was also included. We compared a range of low-passage-number disease and carriage isolates from different serogroups and sequence types from the ESR meningococcal collection (Table 1).

All disease-associated isolates tested inhibited wound healing, compared to uninfected cultures (Fig. 1A; n ≥ 6 for each isolate). In contrast, carriage-associated isolates showed various degrees of wound repair inhibition, ranging from almost complete wound closure to wound repair inhibition similar to (or even greater than) that observed for disease-associated isolates. This likely reflects the genetic diversity among carriage-associated isolates and suggests that, while the ability to inhibit wound repair may contribute to virulence, it is not sufficient to confer virulence, a multifactorial trait, on carriage-associated isolates. One isolate, indistinguishable from the NZ epidemic strain type by standard molecular typing methods but isolated from the nasopharynx of a healthy individual (NZCM238), inhibited wound repair to a similar extent to isolates recovered from disease cases. We found that low-passage-number isolates are essential for this assay, as we observed significant loss of the ability to inhibit wound repair in as few as 10 laboratory subcultures of a disease isolate (data not shown). The related Neisserial species, N. gonorrhoeae and N. lactamica, which rarely cause invasive or disseminated disease, do not inhibit wound repair to the same extent as N. meningitidis (Fig. 1B).

FIG 1.

Inhibition of wound repair by meningococcal isolates. (A) A range of disease- and carriage-associated isolates from our collection were tested for their ability to inhibit wound repair in the 16HBE bronchial epithelial cell line, using an Oris cell migration assay. The graph depicts the remaining wound area as a percentage of the initial area. The disease-associated isolates all inhibit wound repair, while the carriage-associated isolates are highly variable. No significant difference is seen between the disease- and carriage-associated isolates, due to variation among carriage isolates. (P = 0.64 [unpaired Student t test]). (B) The related species, N. lactamica and N. gonorrhoeae, do not inhibit wound repair to the same degree as disease-associated N. meningitidis (NZ98/254). ***, P < 0.0001. (C) Images of the Oris cell migration wound area are shown, including the initial wound area (wound), and the remaining wound areas after 20 h for cells that were uninfected or infected with a disease-associated isolate (NZ98/254) or a carriage-associated isolate (NZCM107). Scale bar, 500 μm. (D) The results were confirmed for a selected N. meningitidis disease- and carriage-associated isolate for primary tracheal cells. *, P < 0.03; **, P < 0.01.

We selected two N. meningitidis isolates for further study: NZ98/254, a disease-associated isolate which belongs to the serogroup B New Zealand epidemic strain type, and NZCM107, a serogroup B carriage-associated isolate. Representative images of 16HBE cells in the Oris assay, after 20 h, are shown in Fig. 1C; cells infected with NZ98/254 exhibited reduced cell migration; neither N. meningitidis strain caused any discernible rounding or detachment.

We next investigated the ability of these two isolates to inhibit wound repair in primary human tracheal epithelial cells (HTEpC; Promocell, Germany). As shown in Fig. 1D, NZ98/254 significantly inhibited wound healing in primary cells compared to uninfected cells (P = 0.0023, Mann-Whitney test, n = 7) and NZCM107-infected cells (P = 0.022, Mann-Whitney test, n = 7). After 20 h, uninfected cultures had ∼16% of the initial wound area left, whereas in NZ98/254-infected cultures, 43% of the initial wound area remained. NZCM107 also inhibited wound healing (28% wound area remaining), although to a lesser extent than for NZ98/254.

Disease-associated N. meningitidis induces actin cables and reduces lamellipodium formation along the wound edge.

Actin cytoskeleton remodeling is required for efficient wound healing. Lamellipodia (ruffles) are membrane protrusions supported by the actin network, located at the front of migrating cells, and a major driving force for cell migration (39). N. meningitidis have been shown to affect the endothelial cell actin cytoskeleton during attachment and invasion (40). To investigate whether N. meningitidis inhibit wound healing of epithelial cells by altering the actin cytoskeleton, we used rhodamine-tagged phalloidin to visualize the actin cytoskeleton in migrating cells. Confluent 16HBE cells were wounded by scratching with a pipette tip and infected with a variety of disease- or carriage-associated N. meningitidis isolates. After 14 h, a time point at which the cells were still actively migrating, cultures were fixed and stained with rhodamine-tagged phalloidin. The wound edge of uninfected cells was lined with lamellipodia (Fig. 2A, arrows), whereas the wound edge for cultures infected with NZ98/254 was lined with bright actin cables (Fig. 2B, triangles). Cultures infected with NZCM107 showed an intermediate morphology, where some cells had lamellipodium-like structures and others had actin cables (Fig. 2C). We calculated the percentage of cells along the wound edge that had lamellipodium-like structures in uninfected, disease-associated, and carriage-associated infected cultures, counting a minimum of 50 cells from two independent experiments for each sample. In uninfected cultures, 57.5% of cells (n = 91 cells) along the wound edge had membrane ruffles, whereas in cultures infected with disease-associated isolates, on average 11.6% of cells (5 strains, n = 462 cells) had these structures. For cultures infected with carriage-associated isolates, we found 19.9% of cells (9 strains, n = 705 cells) had lamellipodium-like structures. When we plotted the percentage of cells with membrane ruffles against wound healing (as a percentage of remaining wound area), we found a significant negative correlation (n = 16, Pearson correlation P = 0.007, R² = 0.42) between remaining wound area and the percentage of cells with ruffles along the wound edge (Fig. 2D). This correlation suggests that N. meningitidis inhibits wound repair by affecting actin cytoskeleton dynamics. Cells away from the wound edge did not display significant morphological differences between infected and uninfected cultures (data not shown).

FIG 2.

Actin dynamics of uninfected cells (A) versus cells infected with the disease-associated isolate (B) or a carriage-associated isolate (C) at 14 h after infection. Actin cables were visualized with rhodamine-tagged phalloidin staining. Lamellipodia (ruffles) are indicated by arrows (A and C), while actin cables are indicated by triangles (B and C). (D) The percentage of the initial wound area after 20 h was plotted against the percentage of infected cells that had lamellipodium-like structures at 14 h, revealing a significant negative correlation.

Inhibition of wound healing is independent of RhoA-GTP and Rac1-GTP levels.

The Rho family of small GTPases are important regulators of actin dynamics, and are required for efficient cell migration (41, 42). To examine whether N. meningitidis infection alters the activation of members of the Rho family of GTPases, we measured the levels of RhoA-GTP and Rac1-GTP in 16HBE cell lysates derived from infected or uninfected cultures, using G-LISA activation assays (Cytoskeleton, Inc.). We chose two different time points for these measurements, including a late time point (14 h), where we observed the actin phenotypes, and an early time point (5 min postinfection). We measured the RhoA-GTP and Rac1-GTP levels in both intact and wounded cell cultures, in the presence or absence of representative disease- and carriage-associated isolates. We found that the RhoA-GTP and Rac1-GTP levels did not significantly differ between uninfected or infected cultures in either wounded or nonwounded conditions at either time point (Fig. 3, standard error of the mean [SEM], n ≥ 3 for each data point).

FIG 3.

Levels of activated RhoA-GTP (A) and Rac1-GTP (B) for cells infected with various N. meningitidis isolates at an early time point (5 min) and a late time point (14 h). A total of 50 μg of total protein was assayed per sample (n ≥ 3 replicates/sample). The “W” indicates the cells were mechanically wounded prior to infection. The negative control (blank) was subtracted from each sample, which was normalized to the positive control. A ratio of 1 corresponds to approximately 3 ng of Rac1-GTP and 1 ng of RhoA-GTP. No significant differences were seen for Rac1 and RhoA activation between the different N. meningitidis isolates or the no-bacteria control.

Inhibition of wound healing is independent of type IV pili.

Many of the reported interactions between Neisseria meningitidis and endothelial cells are mediated by type IV pili (T4P) (13, 14, 43, 44). Morand et al. (45) showed that N. meningitidis expressing PilC1 inhibits cell motility in human ME180 endometrial epithelial cells. PilC1 regulates T4P and facilitates pilus-dependent adhesion to host cells (46, 47). To test whether PilC1 and T4P are involved in wound healing inhibition, we generated mutants of NZ98/254 with deletions in pilC1 (NZ98/254 ΔpilC1), the pilE antigenic variation region, which encodes pilin, the major protein component of T4P (NZ98/254 ΔpilE ΔpilS), and pilM, a small accessory protein required for T4P assembly (NZ98/254 ΔpilM [48]). The ΔpilE ΔpilS mutant and the wild-type parent, NZ98/254, were examined by transmission electron microscopy to confirm the absence of pili in the mutant and their presence in the wild type. We infected wounded 16HBE cells and measured the area of wound remaining after 20 h. As shown in Fig. 4A, there were no significant differences between the deletion mutants and the parent strain in inhibition of wound healing, suggesting that T4P are not required for inhibition of wound healing in 16HBE cells. We did not observe differences in the actin cytoskeleton in cells infected with the wild type compared to a pilin mutant (Fig. 4B). We also observed that the carriage-associated isolate, NZCM107, which does not inhibit wound repair, adheres to 16HBE cells in significantly greater numbers than NZ98/254 (see Fig. S1 in the supplemental material), further evidence that wound repair inhibition does not correlate directly with adherence.

FIG 4.

Oris cell migration assay results demonstrate that wound repair inhibition by the disease-associated wild-type strain is similar to the inhibition by various pilin mutants. (A) Deletion of the genes encoding PilE/PilS (major pilin subunit), PilC1, or PilM (pilin accessory proteins) did not affect the ability of the bacteria to inhibit wound repair in 16HBE cells. No significant differences were seen between the NZ98/254 parent strain and the ΔpilE ΔpilS mutant (P = 0.16), the ΔpilC1 mutant (P = 0.59) or the ΔpilM mutant (P = 0.20 [all unpaired, unequal variance Student t test]). (B) The actin dynamics of 16HBE cells infected with the wild-type (NZ98/254) were similar to those of cells infected with a pilin mutant. Scale bar, 20 μm.

Other N. meningitidis surface-associated virulence factors are dispensable for inhibition of wound repair.

Multiple N. meningitidis surface structures have been shown to be important in meningococcal disease. Many of these are outer membrane proteins that are antigenic or interact with host cells, e.g., PorA and PorB porins, the OmpA family protein (also known as rmp or class 4 outer membrane protein), or adhesins, e.g., OpcA and the adhesion and penetration protein (hap) (43, 49–55). Other surface structures, such as the capsule or lipooligosaccharide (LOS), are associated with more virulent strains of N. meningitidis (56). We therefore generated targeted mutants for genes encoding these proteins or, in the case of the capsule and LOS, proteins involved in their synthesis. The loss of the capsule in the ΔsiaD mutant was confirmed by a slide agglutination test. The LOS mutants targeted the lgtA and lgtC glycosyltransferases. An ΔlgtA mutant lacks the lacto-N-neotetraose α-chain, which has been shown to play a role in the PI3K/Rac1 signaling pathway in endothelial cells (40). Mutants were constructed in the wild-type NZ98/254 background, and we tested the ability of the mutants to inhibit wound repair. However, none of the mutants tested significantly lost the ability to inhibit wound repair, relative to the wild-type parent (results summarized in Table 3). This suggests that either another, untested structure is the bacterial factor that inhibits wound repair or that inhibition of wound repair is a redundant activity among multiple proteins.

TABLE 3.

Summary of wound repair inhibition by mutants

Bacterial strain	Gene function	% Wound area remaining ± SD	No. of samples
Control (no bacteria)	NAa	9 ± 8	22
NZ98/254 strain	NA	45 ± 9	21
ΔporA mutant	Major outer membrane porin	35 ± 4	6
ΔporB mutant	Major outer membrane porin	38 ± 15	7
ΔsiaD mutant	Capsule synthesis	30 ± 4	6
ΔlgtA mutant	Lipooligosaccharide synthesis	48 ± 6	11
ΔlgtC mutant	Lipooligosaccharide synthesis	39 ± 5	6
ΔopcA mutant	Opacity protein A	48 ± 13	6
ΔNZNMB0533_1864 mutant	OmpA family protein	40 ± 8	6
Δhap mutant	Adhesion and penetration protein	40 ± 8	16

^aNA, not applicable.

Inhibition of wound repair requires live bacteria and active translation.

Our data indicate that the capsule, LOS, T4P, other adhesins and the major porins are not required for inhibition of wound repair. We next tested whether the inhibition of wound repair required live bacteria or active protein synthesis. N. meningitidis were killed by heating at 56°C for 1 h or fixed with 2% paraformaldehyde for 20 min. Approximately 100 μl of the killed culture (OD₆₀₀ of 1) was spread onto plates and incubated overnight, to ensure killing was complete. When wounded cultures were infected with killed bacteria, either by heat or fixative, delays in wound closure were not observed for any MOI (Fig. 5A).

FIG 5.

Wound repair inhibition by disease-associated N. meningitidis requires live bacteria undergoing active protein synthesis. (A) N. meningitidis disease isolate NZ98/254 inhibits wound repair, but heat-killed (HK) or paraformaldehyde-fixed (PFA) bacteria are not able to inhibit wound repair, even at fairly high MOI. (B) Similarly, bacteria must be able to actively synthesize protein, since chloramphenicol-treated bacteria (CAM) fail to inhibit wound repair, relative to nontreated NZ98/254. Chloramphenicol alone does not affect the ability of 16HBE cells to migrate. ***, P < 0.0001.

To examine whether active protein synthesis is required for inhibition of wound repair, we analyzed wound closure in the presence of chloramphenicol, which is a potent protein synthesis inhibitor for both Gram-negative and Gram-positive bacteria. N. meningitidis cultures were pretreated with 50 μg/ml chloramphenicol for 30 min at 37°C to halt protein synthesis and then added at various MOIs to wounded 16HBE cells in the presence of the drug. Chloramphenicol alone does not affect 16HBE migration (Fig. 5B, no bacteria plus chloramphenicol control). In contrast to bacteria that were not exposed to chloramphenicol, treated bacteria were unable to inhibit wound repair at MOIs of 10 and 100 (Fig. 5B).

DISCUSSION

The airway epithelium acts as an important physical barrier against pathogen invasion. Although N. meningitidis is frequently found residing harmlessly on nasopharyngeal mucosal surfaces, the integrity of the epithelium helps to ensure that the pathogen does not gain access to deeper tissues. However, on rare occasions, this does occur, though we know little about the mechanism of invasion of deeper host tissues by N. meningitidis.

Any damage to the airway epithelium by environmental factors or inflammation needs to be repaired quickly to stop bacteria from taking advantage of the breach and invading the host. We hypothesized that certain virulent N. meningitidis isolates could take advantage of small wounds by delaying wound healing and crossing the epithelium. Indeed, it has been shown that various environmental factors that disrupt the respiratory epithelium, such as exposure to dust or cigarette smoke, can increase the likelihood of acquiring meningococcal disease. In addition, host genetic polymorphisms that increase the likelihood of acquiring bacterial sepsis following burn wounds also increase the likelihood of acquiring invasive meningococcal disease, without changing the risk of mortality following acquisition of the disease (57, 58). However, these environmental factors generally only raise the risk of infection by more virulent meningococcal isolates, since the risk of invasive disease from carriage-associated isolates remains low.

Other pathogenic bacteria, including Helicobacter pylori and Pseudomonas aeruginosa, have been shown to inhibit epithelial wound repair (21–23). P. aeruginosa inhibits wound repair in the lung epithelium-like cell line A549 and induces cytotoxic effects, including cell rounding and detachment at the wound edge (21). We did not see such effects in cells infected with N. meningitidis, where we observed that cells along the wound edge, as well as confluent cells distal to the wound, appeared healthy with intact junctions (Fig. 1B and 2). Instead, N. meningitidis attenuated wound repair by reducing lamellipodium formation along the wound edge in 16HBE cells. We observed bright actin cables along the wound edge in disease-associated N. meningitidis-infected cultures, whereas in uninfected cultures the wound edge was lined with lamellipodium-like structures. Wound healing in 16HBE cells is driven by collective migration of a continuous sheet. During wound repair the cells do not break away from the wound edge and migrate as individuals. It has been shown that continuous lamellipodia along the wound edge of a sheet of migrating cells partly generate the protrusive force required for pulling the cell sheet in the direction of movement (59).

Actin dynamics are tightly regulated during cell migration and are controlled by a large number of proteins. In particular, the three members of the Rho family of small GTPases—RhoA, Rac1, and CDC42—are central to these processes (41, 42, 60). It is generally accepted that activation of the Rac1 isoform leads to lamellipodium formation, that activation of the RhoA isoform leads to the formation of stress fibers and focal adhesion, and that CDC42 activation leads to the formation of filopodial extensions (39). In N. meningitidis-infected cultures, we observed an increase in actin bundles and a decrease in lamellipodia along the wound edge, a finding reminiscent of the wound edge in 16HBE cells expressing constitutively active Rho-GTP or dominant-negative Rac (42). Therefore, we predicted that N. meningitidis infection would reduce Rac-GTP and increase Rho-GTP levels in the host cell. However, we found no significant differences among isolates tested or between wounded and unwounded cells at early or late time points. This could be due to the fact we were measuring the total amount of Rho/Rac-GTP in the cultures, whereas changes in activation levels were only occurring in cells at the wound edge. Although care was taken to make multiple wounds in the cell monolayer to increase the percentage of cells in the vicinity of a wound edge, there are still large areas of confluent cells distal to wound edges. Activation or inhibition of Rho-GTPases are not even across a wounded culture; rather, their regulation is under tight spatiotemporal control (reviewed by Pertz (61). Regulation of Rho-GTPases by N. meningitidis at the wound edge might be masked by lack of regulation in cells distant from wounds. Similarly, N. meningitidis may be influencing the activity of downstream effectors of the Rho-GTPase pathways. Although several studies have shown that N. meningitidis adhesion and invasion into human endothelial cells is dependent on Rho-GTPase signaling, none have directly measured protein levels after infection (40, 62). The related pathogen, N. gonorrhoeae, has been shown to increase levels of RhoA-GTP in a type IV pilin-dependent manner, resulting in reduced invasion of ME180 endometrial epithelial cells (63). However, in our model, we did not see any differences in cell migration or in actin dynamics in cells infected with pilin mutants.

The N. meningitidis factor responsible for inhibition of wound repair remains unidentified. We tested the requirement of multiple N. meningitidis genes encoding surface factors known to be antigenic or to interact with host cells. All of the tested mutants inhibited wound repair in respiratory epithelial cells to levels similar to the wild type. It has previously been shown that PilC1, a T4P accessory protein that facilitates adhesion to host cells, can impair motility in ME180 epithelial cells (45). However, our evidence suggests that PilC1 is not required for inhibition of wound repair in our system. It is important to note that, while the PilC1 gene for the sequenced genome NZ05/33 (NMBNZ0533_0475) is annotated as a pseudogene, we have previously noted by transcriptomic analysis that the gene is expressed in our cultures of wild-type NZ98/254 (X. Ren and J. Mackichan, unpublished data). The differences observed between our study and that of Morand et al. (45) could be due to differences in the N. meningitidis strain (e.g., the extent of laboratory passage) or host cell type (16HBE bronchial versus ME108 endometrial epithelial cells). In addition, we found mutants in genes involved in other aspects of T4P assembly, including those encoding the major structural protein of T4P (pilE) and a protein required for T4P assembly (pilM) did not reverse the wound repair inhibition phenotype, suggesting that T4P are not required. Similarly, mutagenesis of other N. meningitidis adhesins and major porins, including OpcA, the OmpA family protein, the adhesion and penetration autotransporter (hap), PorA, and PorB, did not reverse the inhibition phenotype. Genes required for capsule and LOS biosynthesis were similarly dispensable. The possible role of the Opa adhesins was not addressed, since they are present in multiple copies in the genome and could not all be deleted (43).

By infecting cells with dead bacteria or in the presence of a translation inhibitor, we have shown that live bacteria and new protein synthesis are required to inhibit wound healing. Killing by heat or fixing by paraformaldehyde preserves the external structures of the bacteria, suggesting that stable physical external structures are not involved in attenuation of wound repair. This is also supported by our results following the mutagenesis of genes encoding components of major external structures. The requirement of active protein synthesis suggests that the inhibitory factor may be synthesized in response to the presence of host cells. Together, these data suggest the possibility of an uncharacterized protein surface structure involved in the manipulation of host processes, potentially representing a novel vaccine candidate. An alternative explanation is that the bacteria are consuming a nutrient in the cell culture supernatant that is required for cell migration. As additional N. meningitidis genomes have been sequenced and published, multiple candidate genes have been identified, including potential toxins and secreted or surface proteins, such as two-partner secretion systems (56); these candidates are yet to be fully characterized and may be responsible for the inhibition of host cell migration. Further efforts are ongoing and are concentrated on identification of the N. meningitidis wound repair inhibition factor through screens and targeted deletion of additional candidate genes.