ABSTRACT
Meningococcal internalization into human cells is likely to be a consequence of meningococcal adhesion to human epithelial and endothelial cells. Here, we identified three transposon mutants of Neisseria meningitidis that were primarily defective in the internalization of human brain microvascular endothelial cells (HBMEC), with insertions occurring in the gltT (a sodium-independent l-glutamate transporter) gene or its neighboring gene, NMB1964 (unknown function). NMB1964 was tentatively named gltM in this study because of the presence of a mammalian cell entry (MCE)-related domain in the deduced amino acid sequences. The null ΔgltT-ΔgltM N. meningitidis mutant was also defective in the internalization into human umbilical vein endothelial cells and the human lung carcinoma epithelial cell line A549, and the defect was suppressed by transcomplementation of the mutants with gltT + -gltM + genes. The intracellular survival of the ΔgltT-ΔgltM mutant in HBMEC was not largely different from that of the wild-type strain under our experimental conditions. Introduction of a1-bp deletion and amber or ochre mutations in gltT-gltM genes resulted in the loss of efficient internalization into HBMEC. The defect in meningococcal internalization into HBMEC and l-glutamate uptake in the ΔgltT-ΔgltM mutant were suppressed only in strains expressing both GltT and GltM proteins. The efficiency of meningococcal invasion to HBMEC decreased under l-glutamate-depleted conditions. Furthermore, ezrin, a key membrane-cytoskeleton linker, accumulated beneath colonies of the gltT + -gltM + N. meningitidis strain but not of the ΔgltT-ΔgltM mutant. These findings suggest that l-glutamate influx via the GltT-GltM l-glutamate ABC transporter serves as a cue for N. meningitidis internalization into host cells.
Neisseria meningitidis (meningococcus) is a Gram-negative diplococcus that survives in humans only by colonizing the nasopharynx. N. meningitidis affects human populations worldwide, and in some cases it can spread into the bloodstream, where it causes septicemia and, furthermore, induces meningitis when it passes through the blood-brain barrier and reaches the cerebrospinal fluid (CSF) (6, 44, 56). While many molecules involved in meningococcal pathogenesis have been identified so far, pili and the outer membrane proteins Opc and Opa are the most characterized virulence factors in N. meningitidis (reviewed in references 6 and 56). It is reported that meningococcal pili bind to CD46 in human cells for meningococcal infection (18, 19) although pili-CD46 binding is controversial (6, 20). It was also revealed that Opa protein binds to carcinoembryonic antigen-related cellular adhesion molecule 1 (CEACAM1) and CEACAM3 in human cells (reviewed in reference 56). Opc proteins primarily act as both adhesive and invasive factors of human peripheral endothelial and epithelial cells (47, 58-61). It is also reported that Opc functions solely as an invasive factor in the context of infection to human brain microvascular endothelial cells (HBMEC) via human fibronectin binding (53), but this idea is controversial (34).
Many bacterial genomes have been sequenced by recent low-cost and efficient sequencing methods, and many neisserial whole-genome sequences have been reported (reviewed in reference 37). Genome mining to investigate meningococcal autotransporters from whole-genome sequences identified several minor adhesion proteins: App (39), NhhA (36, 40), MspA (52), and TspA (30). In addition, NadA has also been identified as an adhesion and invasion factor in human epithelial cells (5). However, significant pathogenicity islands have not been identified so far (56). While a prominent virulence system, such as a type III secretion system (T3SS), has not been found in N. meningitidis, N. meningitidis strains can be efficiently internalized into human endothelial and epithelial cells in a nonencapsulated state (57, 59). These results imply that, in contrast to pathogenic enterobacteria such as enterohemorrhagic Escherichia coli, unidentified pathogenic factors are dispersed among the meningococcal genomes and play a role in meningococcal pathogenesis.
In addition, putative open reading frames (ORFs) have been identified by genome sequence analyses in bacteria, and putative ORFs with unknown function have been recorded in the data banks. Among them, the deduced amino acid sequences with some similarity to mycobacterial mammalian cell entry (MCE) proteins are defined as belonging to the “MCE-related” or “MCE family” domain, which is widely found in bacterial genomes (EMBL-EBI database [http://ebi.ac.uk]). MCE proteins in Mycobacterium tuberculosis (the etiologic agent of tuberculosis) are identified as invasion factors while the biological functions of the MCE-related proteins remain unclear (24, 35, 38).
It is well known that bacterial infection elicits an alternation of signaling pathways concomitant with cytoskeletal reorganization in the host cells (21). Ezrin, radixin, and moesion (ERM) proteins are a family of widely distributed membrane-associated proteins responsible for linking the plasma membrane to the underlying actin cytoskeleton (4). Ezrin is one of the key molecules controlling both cell shape and signaling (51). It has been shown that pathogenic bacteria induce ERM accumulation beneath the bacterial colony and that this accumulation is required for bacterial internalization into host cells (32, 41, 62). It has also been shown in Neisseria that ezrin is accumulated beneath the bacterial colony upon infection with Neisseria gonorrhoeae and N. meningitidis (12, 27).
In a previous study, we found that N. meningitidis strains of sequence type 2032 (ST-2032), in which opc gene was absent, were highly infectious to human endothelial and epithelial cells in vitro in capsule-negative (ΔsiaB-ΔsiaD) genetic backgrounds (45, 47). To further study the in vitro high infectious ability of the ST-2032 N. meningitidis strain in an Opc-independent manner, we performed signature tag mutagenesis (STM) against the ΔsiaB-ΔsiaD ST-2032 N. meningitidis strain (2, 17; also unpublished data). While characterizing the mutants, it was found that three mutants were disrupted by a transposon in two neighboring genes. Two of three mutants (44C7 and 52C6) were disrupted in the gltT (NMB1965) gene, and the other (18H7) was disrupted in the neighboring gene NMB1964, in which the MCE-related domain was found, tentatively named gltM (GltT-associated MCE-related gene) in this study. These transposon mutants were primarily defective in invasive activity in HBMEC. Here, we examined the possible MCE-like function of GltM in meningococcal infection of human cells and also studied the relationship between meningococcal l-glutamate uptake via the GltT-GltM l-glutamate ABC transporter and meningococcal pathogenesis.
MATERIALS AND METHODS
Bacterial growth conditions.N. meningitidis strains stored at −80°C were routinely grown on GC agar plate at 37°C in 5% CO2 (45). To select kanamycin-resistant N. meningitidis strains, brain heart infusion (Becton-Dickinson) agar containing 3% defibrinated horse blood (Nihon Biotest, Japan) was used. E. coli strains were grown on L plate or in L broth liquid culture at 37 or 30°C. When required, antibiotics were added at the following concentrations: kanamycin at 150 μg/ml, chloramphenicol at 5 μg/ml, erythromycin at 4 μg/ml, and spectinomycin at 75 μg/ml for N. meningitidis; kanamycin at 50 μg/ml, ampicillin at 50 μg/ml, chloramphenicol at 10 μg/ml, erythromycin at 150 μg/ml, and spectinomycin at 75 μg/ml for E. coli.
All the meningococcal strains used in this study were derivatives of the N. meningitidis HT1125 strain. The HT1125 strain is an unencapsulated (ΔsiaB-ΔsiaD::kan) mutant of an NIID280 N. meningitidis strain, which is untypeable, Opc− Pili+ Opa+, and belongs to the ST-2032 complex (47). All of the strains used in this study are listed in Table 1.
Strains used in this study
Tissue culture.HBMEC, human umbilical vein endothelial cells (HUVEC), and the human lung carcinoma epithelial cell line A549 were cultivated as described previously (45, 47).
Determination of associated and internalized bacteria in host cells.Infection assays using tissue culture were performed as described previously (45) and by Cunha et al. (10) with some modifications. Bacteria suspended in MCDB131 medium with 10% fetal bovine serum (FBS), 90 μg ml−1 heparin, and 3 mM glutamine (assay medium [AM]) was used at a multiplicity of infection (MOI) of 500. A higher MOI was used for an infection study to obtain a high level of internalized bacteria (34). For an infection assay under l-glutamate-limited conditions, MCDB131 lacking l-glutamic acid (specially manufactured by Funakoshi, Japan) and without l-glutamine and FBS supplementation was used as the AM, and the bacteria and the human cultured cell monolayers were washed five times with phosphate-buffered saline (PBS) prior to the infection assay. Human cell monolayers grown on gelatin-coated 96-well tissue culture plates (IWAKI) at 37°C in 5% CO2 for 2 days were infected for 4 h. After removal of nonadherent bacteria by washing, cell-associated bacteria were released with saponin treatment and enumerated by counting viable cells as described previously (45, 47). Bacterial invasion was determined by a gentamicin protection assay (45, 47). All samples were tested in duplicate, and experiments were repeated at least four times.
Production of anti-GltM protein rabbit antiserum.A 380-bp DNA fragment containing a putative hydrophilic domain of GltM (see Fig. S1 in the supplemental material) was amplified with a set of primers (NMB1964-pET303-1 and NMB1964-pET303-2), which resulted in deletion of the N-terminal 34 amino acid residues. The DNA fragment was digested with NsiI and XhoI and cloned into the same cutting sites of the expression vector pET303CT-His (Invitrogen), resulting in pHT669. The plasmid pHT669 was transformed into E. coli strain C41 (DE3) (CosmoBio, Japan), and the transformant was cultured in 250 ml of MagicMedia (Invitrogen) at 30°C overnight with shaking. The subsequent purification of a recombinant protein and generation of a polyclonal rabbit serum to a putative hydrophilic domain of the GltM protein were preformed as described previously (48).
Construction of meningococcal mutants.An N. meningitidis mutant with a deletion of the gltT-gltM genes was constructed as follows; a 0.7-kb DNA fragment containing the upstream region of the gltT gene was amplified with a set of primers (NMB1964+NMB1965-1 and NMB1965-2) from HT1125 chromosomal DNA and cloned in pGEM-T easy (Promega) to construct pH 635. A 1-kb DNA fragment containing the downstream region of the gltM gene was also amplified with a set of primers (NMB1964+NMB1965-2 and NMB1964-1) and cloned in pGEM-T easy to construct pHT638. The 0.7-kb SphI-HpaI DNA fragment from pHT635 was inserted into the same sites of pHT638 to generate pHT641. A 0.8-kb fragment containing an erythromycin-resistance gene (ermC) was inserted into an HpaI site of pHT641, resulting in pHT644. Five hundred nanograms of linearized pHT644 was transformed into HT1125, and an erythromycin-resistant (Ermr) clone was selected, resulting in a gltT-gltM deletion mutant, named HT1414.
A ΔgltT-ΔgltM mutant ectopically complemented with wild type gltT+-gltM+ genes at the ggt locus was constructed as follows: a 3-kb DNA fragment containing the gltT+-gltM+ genes was amplified with a set of primers (NMB1964+NMB1965 −1 and NMB1964+NMB1965 −2) from HT1125 chromosomal DNA and cloned in pGEM-T easy (Promega) to construct pHT628. A 1.2-kb fragment containing a chloramphenicol resistance gene (cat) was inserted in a blunted SpeI site of pHT628, resulting in pHT631. The blunted NotI-NotI fragment containing the gltT+-gltM+-cat genes was inserted in a blunted BstXI site, which is located in the middle of the coding region of ggt of pHT195 (48) to construct pHT634. Five hundred nanograms of linearized pHT634 was transformed into HT1414, and chloramphenicol-resistant (Cmr) clones were selected, resulting in the ectopic complementation of gltT+-gltM+ genes at ggt loci in the ΔgltT-ΔgltM mutant, named HT1428.
The derivatives of gltT-gltM mutants including 1-bp deletions, gltT with an amber mutation [gltT(Am)], and gltM with an ochre mutation [gltM(Oc)] were constructed using a PrimeSTAR Mutagenesis Basal Kit (Takara Bio, Japan) with the respective primer set (see Table S1 in the supplemental material) and pHT634 as template DNA. The obtained plasmid (see Table S2) was linearized and transformed into strain HT1414. The resultant Cmr transformants were selected for an N. meningitidis mutant that was subjected to site-directed mutagenesis (see Table 1).
The meningococcal mutant expressing the gltM+ gene under the control of the opc promoter (popc) at the iga locus was constructed as follows: a 375-bp DNA fragment containing the opc promoter and Shine-Dalgarno sequences was amplified with a set of primers (opc-21 and opc-B) from the chromosomal DNA of the H114/90 strain, in which the Opc protein is expressed in great abundance (47). A 1.4-kb DNA fragment containing the coding region of the gltM gene was also amplified with a set of primers (NMB1964-B and NMB1965-2) using pHT628 as template DNA. Aliquots of the above two DNA fragments were mixed in a 50-μl PCR mixture without DNA polymerase and primers and heated at 94°C, followed by gradual cooling to 37°C overnight. A second PCR was performed by adding 10-pmol primers (opc-21 and NMB1965-2) and PrimeSTAR GXL DNA polymerase (Takara Bio). The 1.8-kb-amplified PCR fragment was phosphorylated and cloned into the HincII site of pTWV228 (Takara Bio, Japan) to generate pHT751. A 1-kb SmaI-HincII spectinomycin resistance gene (spc) was inserted into the blunted SphI site of pHT751, which is in the multicloning site, resulting in pHT766. A 2.8-kb DNA fragment containing the popc-gltM+-spc genes was amplified with a set of primers (M13-RV-iga-5′ and M13-RV-iga-3′) using pHT766 as a template by PrimeSTAR GXL DNA polymerase. The DNA fragment was cloned into the EcoT22I site of pHT475, which has the iga gene of NIID280 N. meningitidis strain in pBR322, using an In-Fusion Advance PCR cloning kit (Clontech). The resultant plasmid named pHT769 was linearized and transformed into N. meningitidis strains HT1452, HT1453, and HT1414, and spectinomycin-resistant (Spcr) clones were selected, resulting in the ectopic complementation of the popc-gltM+-spc genes at iga loci in strains HT1539, HT1533, and HT1538, respectively.
Western blotting.Western blotting was performed as described previously (45).
Preparation of cellular protein fractions.Cellular fractionation was performed as described previously (45).
Monitoring meningococcal survival in host cells.The meningococcal intracellular survival assay in HBMEC was performed as follows: confluent HBMEC monolayers seeded on gelatin-coated 96-well tissue culture plates (IWAKI) were infected with meningococcus at an MOI of 5,000 for 2 h to obtain a maximal number of internalized bacteria (45). The high dosage of meningococci did not show the cytotoxic effect on HBMEC (data not shown). The monolayers were washed with AM four times and then incubated with AM containing 150 μg ml−1 gentamicin for 1 h to kill extracellular bacteria. Gentamicin-containing AM was replaced with AM, and samples were further incubated at 37°C in 5% CO2 to monitor the number of bacteria inside HBMEC. This time point was defined as zero hour. Intracellular bacteria were determined at time zero and at 1, 2, and 4 h, and the bacterial numbers as number of CFU were determined by plating on GC agar plates after appropriate dilution.
Bacterial l-glutamate uptake assay.An l-glutamate transport assay was performed as described previously (28) with some modification. N. meningitidis strains grown overnight on GC agar plates at 37°C in 5% CO2 were suspended in 1 ml of buffer A (50 mM potassium buffer, pH 6.9, and 0.5 mM MgCl2) and harvested by centrifugation at 10,000 × g for 2 min at 4°C. The bacterial pellet was resuspended in 1 ml of buffer A and centrifuged again. The procedures were repeated five times to wash the bacteria. The resultant pellets were suspended in buffer B (buffer A containing 300 μg ml−1 chloramphenicol) to adjust the bacterial concentration at an optical density at 600 nm (OD600) of 8. Twenty microliters of bacterial suspension was mixed with 20 μl of buffer A containing 3% glycerol and 60 mM NaCl. The assays were initiated by the addition of 20 μl of buffer A containing 1.5 μM l-[3,4-3H]glutamic acid (specific activity, 1.83 TBq/mmol) (Perkin-Elmer Life Science), incubated at room temperature for 20 s, and then stopped by vacuum filtration through a 13-mm-diameter 0.45-μm-pore-size HAWP membrane (Millipore). The filters were washed with 1 ml of buffer A and dried. Radioactivity was determined by scintillation counting in an 8-ml vial containing 5 ml of Filtron-X (National Diagnostics) after incubation for 15 min. Background binding of l-[3,4-3H]glutamic acid to membrane or bacteria was determined in control samples lacking bacteria or glycerol, and this value was subtracted from all values. The protein concentration was determined by a Protein Quantification Kit-Rapid (Dojindo, Japan), with bovine serum albumin (BSA) as a standard. l-Glutamate transport values are expressed as nanomoles of substrate transported per min per mg of bacterial protein by the means ± standard deviations (SD).
Observation of ezrin accumulation and actin filament by immunofluorescence staining.HBMEC were grown on an eight-well Lab-Tek II chamber slide at 37°C in 5% CO2 for 2 days. The confluent HBMEC monolayers were washed two times with PBS and then incubated with serum-free AM [AM(−S)] at 37°C in 5% CO2 for 16 h. The HBMEC monolayers were infected with meningococcus in AM(−S) at an MOI of 500 for 4 h. The monolayer was washed four times with AM(−S), followed by fixing with 4% paraformaldehyde in PBS for 15 min. The fixed cells were washed with PBS, followed by incubation with 250 μl of 50 mM NH4Cl2 in PBS for 10 min to inactivate residual paraformaldehyde. The monolayers were permeabilized with 0.2% Triton X-100 in PBS, followed by blocking with 2% BSA in PBS. The resultant monolayers were incubated with 100-fold diluted anti-ezrin monoclonal antibody 3C12 (Santa Cruz Biotechnology, Santa Cruz, CA) for 30 min and with a 2,000-fold diluted Alexa Fluor 488-conjugated F(ab′) fragment of rabbit anti-mouse IgG (Invitrogen) under moist and dark conditions for 30 min. To visualize actin filaments, the fixed and blocked monolayers were incubated with 100-fold diluted Alexa Fluor 568-conjugated phalloidin (Invitrogen).
Statistical analyses.Results are expressed as the means ± SD. The percentages of intracellular bacterial survival and l-glutamate uptake were compared using a Student's t test and were considered significant at a P value of <0.05.
RESULTS
Isolation of transposon mutants primarily defective in HBMEC invasion.To further study the in vitro infectious abilities of the HT1125 N. meningitidis strain in an Opc-independent manner, we performed STM and isolated three transposon mutants defective in their abilities to infect HBMEC. Two of three mutants (44C7 and 52C6) were disrupted in the gltT (NMB1965) gene, and the other (18H7) was disrupted in the neighboring gene, NMB1964 (Fig. 1 A; see also Fig. S1 in the supplemental material). While GltT was identified as a sodium-independent glutamate transporter (28), the function of NMB1964 is unknown. BLAST analysis revealed that the deduced protein sequence of the NMB1964 gene had approximately 25% identity and 76% similarity to the N-terminal region of the mycobacterium Mce1a protein (Fig. 1B). In this study, NMB1964 was tentatively named gltM (GltT-associated MCE-related gene).
Mutation in the gltT-gltM genes affected N. meningitidis internalization into human endothelial and epithelial cells. (A) Schematic representation of insertion and deletion mutations in gltT-gltM locus and ectopic complementation of gltT+-gltM+ genes at the ggt locus in N. meningitidis strains. (B) Lipman-Pearson alignment between the deduced amino acid sequence of NMB1964 (named GltM in this study) and the N-terminal part of the mammalian cell entry protein (Mce1A) from the M. tuberculosis H37Rv strain. Identity (*) and synonymous substitution (•) are shown. (C and D) Adherence and internalization of gltT-gltM N. meningitidis mutants in HBMEC and effect of complementation of gltT+-gltM+ genes in the gltT-gltM mutants on bacterial infection. (E and F) Adherence and internalization of gltT-gltM mutants in HBMEC, HUVEC, and the A549 cell line. Bacterial number is represented as CFU. Internalized bacteria were determined as the gentamicin-resistant bacterial number. Ranges represent the mean CFU of adherent bacteria (C and E) and internalized bacteria (D and F) per 104 cells in at least four experiments, and error bars represent the standard error of the means. Graphs in panels C and D show the bacterial number of the N. meningitidis gltT+-gltM+ strain (HT1125), the original gltT or gltM insertion mutants (44C7, 52C6, and 18H7), the backcrossed mutants (HT1388, HT1390, and HT1391), and the ΔgltT-ΔgltM::ermC (HT1414) and ΔgltT-ΔgltM mutants in which the gltT+-gltM+ genes were ectopically complemented (HT1409, HT1410, HT1411, and HT1428) (see Table 1). In panels E and F graphs indicate the bacterial number of the N. meningitidis gltT+-gltM+ (NIID280 and HT1125; open bars) and ΔgltT-ΔgltM::ermC mutant (HT1412 and HT1414; filled bars) strains (Table 1).
To confirm the relationship between the insertional mutations and infection defects in the three mutants, we constructed backcrossed mutants and complemented strains with wild-type gltT+-gltM+ genes in trans at the ggt locus (Fig. 1A) (46) and examined their infectious activities in HBMEC (Fig. 1C and D). The growth rate in assay medium (AM) (see Materials and Methods) and the susceptibility to gentamicin were not different between the gltT-gltM mutants and the gltT+-gltM+ strain HT1125 (data not shown). While the three insertional mutants and the backcrossed mutants adhered to HBMEC a little less efficiently than the gltT+-gltM+ strain HT1125 (Fig. 1C), the number of internalized bacteria decreased to approximately 1/50 of the level of HT1125 (Fig. 1D). The gltT-gltM null mutant HT1414 (HT1125 ΔgltT-ΔgltM::ermC) also was less invasive in HBMEC than the gltT+-gltM+ strain, and complementation with wild-type gltT+-gltM+ genes in trans suppressed invasion defectiveness in all the gltT-gltM mutants (Fig. 1D). The internalization defect of gltT-gltM mutants was also confirmed by the double immunofluorescence technique (16) (data not shown). The invasion defectiveness in gltT-gltM mutants was also observed with infection of human umbilical vein endothelial cells (HUVEC) and the human lung carcinoma epithelial cell line A549 (Fig. 1E and F). These results suggested that the mutation in gltT-gltM genes affected meningococcal internalization much more than adhesion to human cultured endothelial and epithelial cells. In addition, there seemed to be no difference in meningococcal infectious behavior among the human endothelial and epithelial cells so that HBMEC were mainly used for all the following infection experiments in this study.
We also examined the effect of gltT-gltM mutation in an sia+ genetic background (Fig. 1E and F). The wild-type strain NIID280 and the ΔgltT-ΔgltM::ermC mutant HT1412 adhered to HBMEC as efficiently as the isogenic ΔsiaB-ΔsiaD::kan mutants. However, NIID280 was approximately 30-fold less efficiently internalized into HBMEC than the ΔsiaB-ΔsiaD::kan strain HT1125, but the effect of the gltT-gltM mutation on meningococcal internalization in the sia+ genetic background was smaller (Fig. 1F). The results implied that meningococcal internalization was compromised by the presence of capsule synthetic genes. Hereinafter, ΔsiaB-ΔsiaD::kan N. meningitidis strain HT1125 was used as a wild-type strain to focus on internalization into human cultured cells in this study.
Invasion defect in gltT-gltM mutants was not related to the reduction of intracellular survival in HBMEC under our experimental conditions.Since l-glutamate uptake via GltT is required for meningococcal growth and survival in HeLa cells (28) and in mice (9), our finding of an invasion defect in gltT-gltM mutants may be explained by the bacterial growth defect in the cells after internalization. To examine this possibility, meningococcal survival in HBMEC was determined during a 4-h incubation under our experimental conditions (see Materials and Methods). The number of intracellular bacteria of both gltT+-gltM+ and the ΔgltT-ΔgltM N. meningitidis strains did not increase but rather decreased during the 4-h incubation (Fig. 2). The rate of decrease (90% of the number at time zero) in the ΔgltT-ΔgltM mutant was slightly more than that (60% of the zero-time number) of the gltT+-gltM+ strain, but the difference was much less than that in the internalization assay (more than a 50-fold difference). These results suggested that the gltT-gltM mutation affected meningococcal internalization into HBMEC, which is independent of intracellular survival in HBMEC (see Discussion).
Percent survival of intracellular bacteria after the killing of extracellular bacteria by gentamicin (Gen). The upper panel is a diagram of the protocol used to monitor the number of intracellular bacteria after a 2-h infection. Detailed of the procedures are given in Materials and Methods. In the lower panel, percent survival of intracellular bacteria in HBMEC is shown. Percent survival is calculated as follows: (number of CFU at the indicated time/number of CFU at removal of gentamicin) × 100. The percent survival of the gltT+-gltM+ (HT1125) and ΔgltT-ΔgltM (HT1414) N. meningitidis strains is shown. Ranges represent the mean percent survival in at least five experiments, and error bars represent the standard error of the means. #, P ≤ 0.08; *, P = 0.020 (compared to the gltT+-gltM+ strain) .
Localization of GltM in N. meningitidis.BLAST analysis revealed that GltM was highly homologous to the putative ORFs containing the MCE-related domain (data not shown) and to the N-terminal region of mycobacterial Mce1a protein (Fig. 1B). It could be predicted that GltM protein directly acted on HBMEC as an invasion factor, like mycobacterial Mce proteins (8, 24), if GltM was localized on the meningococcal surface. To examine the location of meningococcal GltM protein in N. meningitidis, the cellular proteins were fractionated and analyzed by Western blotting with an anti-GltM rabbit antiserum (see Fig. S2A in the supplemental material). GltM protein was detected only in the outer membrane fraction, where the outer membrane protein Opa was also detected as a control (see Fig. S2A). Moreover, immunofluorescent staining of N. meningitidis strains with an anti-GltM rabbit antiserum revealed that the C-terminal putative hydrophilic domain of GltM was exposed on the meningococcal surface (see Fig. S2B). These results suggested that the GltM protein was located in the meningococcal outer membrane and exposed to the surface.
On the other hand, while we were successful in the preparation of anti-GltT rabbit serum that reacted with recombinant GltT on polyvinylidene difluoride (PVDF) membranes, GltT protein was hardly detected by Western blotting of N. meningitidis extracts (data not shown). These results suggested that the l-glutamate transporter GltT might be a minor protein in N. meningitidis (see Discussion).
Mutations in gltT and gltM genes decreased meningococcal invasion ability in HBMEC.To precisely identify the responsible gene in the gltT-gltM locus for meningococcal internalization, a series of 1-bp deletion mutants, in which a 1-bp deletion mutation was systematically introduced into the gltT-gltM locus at the ggt locus at an interval of approximately 300-bp lengths, was constructed (Fig. 3A; see also Fig. S1 in the supplemental material), and mutants were examined by infection assay with HBMEC. Mutants HT1506, HT1507, HT1508, and HT1509 (representing deletion mutants Δ4, Δ5, Δ6, and Δ7, respectively, in Fig. 3), in which either the gltT or gltM gene was mutated, were defective in HBMEC invasion (Fig. 3B and C). The involvement of gltT-gltM genes in meningococcal internalization was also confirmed by gltT(Am)and gltM(Oc)mutants (HT1452 and HT1453) (Fig. 4A, B, and C; see also Fig. S1). The results suggest that the gltT and gltM genes, but not the truncated NMB1966 or NMB1963 gene, were responsible for meningococcal internalization into HBMEC.
Identification of the responsible genes in the gltT-gltM locus for meningococcal internalization. (A) Schematic representation of 1-bp deletion mutations in gltT-gltM locus. A 1-bp deletion was systematically introduced in gltT-gltM locus, which was ectopically complemented in the HT1414 N. meningitidis strain. The exact position of the deleted nucleotide is shown in Fig. S1 in the supplemental material. (B and C) Adherence and internalization of the N. meningitidis mutants at the gltT-gltM locus in HBMEC. Bacterial number is represented as CFU. Internalized bacteria were determined as the gentamicin-resistant bacterial number. Ranges represent the mean CFU of adherent bacteria (B) or internalized bacteria (C) per 104 cells in at least four experiments, and error bars represent the standard errors of the means. Graphs in panels B and C indicate the bacterial number of the N. meningitidis gltT+-gltM+ strain(HT1125), the ΔgltT-ΔgltM::ermC mutant (HT1414), the ΔgltT-ΔgltM mutant in which gltT+-gltM+ genes were ectopically complemented (HT1428), and the 1-bp deletion mutants (HT1503 to HT1512) (Table 1). (D) Western blotting for GltM proteins. Bacterial extracts equivalent to an OD600 of 0.05 were analyzed by Western blotting. Lane 1, HT1428 (gltT+-gltM+); lane 2, HT1503 (Δ1 mutant) (Fig. 3A); lane 3, HT1504 (Δ2); lane 4, HT1505 (Δ3); lane 5, HT1506 (Δ4); lane 6, HT1507 (Δ5); lane 7, HT1508 (Δ6); lane 8, HT1509 (Δ7); lane 9, HT1510 (Δ8); lane 10, HT1511 (Δ9); lane 11, HT1512 (Δ10).
Effect of amber and ochre mutations on gltT-gltM genes and additional complementation of the gltM gene alone in trans for meningococcal internalization. (A) Schematic representation of amber and ochre mutations in gltT-gltM genes and the additional complementation of the gltM gene under the opc promoter (popc-gltM+) (see Materials and Methods) in N. meningitidis mutants. The exact positions of the amber and ochre mutations are shown in Fig. S1A in the supplemental material. (B and C) Adherence and internalization of N. meningitidis mutants at the gltT-gltM locus in HBMEC. Bacterial number is represented as CFU. Internalized bacteria were determined as the gentamicin-resistant bacterial number. Ranges represent the mean CFU of adherent bacteria or internalized bacteria per 104 cells in at least four experiments, and error bars represent the standard errors of the means. Graphs indicate the bacterial number of the N. meningitidis gltT+-gltM+ strain(HT1125), ΔgltT-ΔgltM mutant (HT1414), the ΔgltT-ΔgltM mutant in which gltT+-gltM+ genes were ectopically complemented (HT1428), the gltT(Am) mutant (HT1452), the gltM(Oc) mutant (HT1453), and the derivatives of additional complementation of the popc-gltM+ gene at the iga locus of the above mutants (HT1539, HT1533, and HT1538) (Table 1). (D) Western blotting for GltM proteins. Bacterial extracts equivalent to an OD600 of 0.05 were analyzed by Western blotting. Lane 1, HT1125 (gltT+-gltM+); lane 2, HT1414 (ΔgltT-ΔgltM); lane 3, HT1428 (ΔgltT-ΔgltM ggt::gltT+-gltM+); lane 4, HT1452 [ΔgltT-ΔgltM ggt::gltT(Am)-gltM+]; lane 5, HT1453 [ΔgltT-ΔgltM ggt::gltT+-gltM(Oc)]; lane 6, HT1539 [ΔgltT-ΔgltM ggt::gltT(Am)-gltM+ iga::popc-gltM+]; lane 7, HT1533 [ΔgltT-ΔgltM ggt::gltT+-gltM(Oc) iga::popc-gltM+]; lane 8, HT1508 (ΔgltT-ΔgltM iga::popc-gltM+). Black and gray arrows show the GltM protein and the degraded product in N. meningitidis, respectively. (E) l-Glutamate uptake into gltT-gltM N. meningitidis mutants. The assay was performed as described in Materials and Methods in the presence of 0.5 μM l-[3,4-3H]glutamate and a final NaCl concentration of 20 mM. *, P < 0.05 compared to the gltT+-gltM+ strain.
However, it was shown by Western blotting that the GltM protein was hardly detected even in the gltT mutants HT1506 (Fig. 3D, lane 5) HT1507 (Fig. 3D, lane 6) and gltT(Am), HT1452 (Fig. 4D, lane 4), as well as being absent in the gltM mutants HT1508 (Fig. 3D, lane 7) HT1509 (Fig. 3D, lane 8) and the gltM(Oc) strain, HT1453 (Fig. 4D, lane 5). It was also confirmed by reverse transcription-PCR (RT-PCR) that the transcription of the gltT-gltM genes was not affected by the gltT and gltM mutations (data not shown). These results implied that GltM protein was degraded in the absence of total balanced amounts of both GltT and GltM proteins in N. meningitidis (see Discussion). In addition, the results raised the question of whether GltM protein alone or both GltT and GltM proteins are required for meningococcal invasion since expression of the GltM protein in the gltT mutant was similar to that in the gltM mutant.
An N. meningitidis mutant in which GltM but not GltT was expressed was not internalized into HBMEC.To further examine the role of GltM and GltT proteins in meningococcal internalization, the gltM+ gene expressed solely by the opc promoter (popc-gltM+) (see Materials and Methods) was further complemented in gltT-gltM mutants at the iga locus (strains HT1539, HT1533, and HT1538) (Fig. 4A), and the ability to infect HBMEC was examined. Complementation with the popc-gltM+ gene in the ΔgltT-ΔgltM mutants did not suppress meningococcal internalization or GltM expression (HT1538) (Fig. 4B, C, and D), which seemed to support the concept that GltM was not expressed in the absence of GltT. On the other hand, the defect in invasion and GltM expression in the gltM(Oc)mutant complemented with popc-gltM+ was suppressed in HT1533 (Fig. 4B, C, and D), suggesting that the gltT gene was actively expressed in gltM mutants. Under the same conditions, while complementation of the popc-gltM+ gene in the gltT(Am) mutant (HT1539) suppressed GltM expression (Fig. 4D), efficient meningococcal internalization into HBMEC was not observed in the HT1539 N. meningitidis strain (Fig. 4C). This result strongly suggested that both GltT and GltM proteins are required for meningococcal internalization rather than GltM protein alone (see Discussion).
l-Glutamate uptake via the GltT transport system in gltT and gltM mutants.While it was clear that both GltT and GltM proteins were required for efficient internalization of HBMEC, the function of the GltM protein in l-glutamate transport via the GltT ABC transport system still remained unclear. We next examined l-glutamate uptake activity via the GltT transport system in a buffer containing 20 mM NaCl, a condition suitable for l-glutamate uptake via the GltT transporter in N. meningitidis (28). Under this condition, l-glutamate was imported 4-fold less efficiently in the ΔgltT-ΔgltM mutant HT1414 than in the gltT+-gltM+ strain HT1125, and the low efficiency of l-glutamate uptake was restored in the HT1428 N. meningitidis strain (Fig. 4E). The low efficiency of l-glutamate uptake was also observed in gltT(Am) and gltM(Oc)mutants, HT1452 and HT1453, respectively (Fig. 4E). Complementation with the popc-gltM+ gene suppressed the low efficiency of l-glutamate uptake in the gltM(Oc)mutant (HT1533) but not in the gltT(Am) mutant (HT1539) or in the ΔgltT-ΔgltM mutant (HT1538) (Fig. 4E). These results implied that both GltT and GltM are required for efficient l-glutamate uptake in N. meningitidis and that GltM also participated in l-glutamate uptake via the GltT transporter (see Discussion). Furthermore, a close relationship was found between meningococcal invasion into HBMEC and l-glutamate influx into N. meningitidis via the GltT-GltM l-glutamate transport system (Fig. 4C and E). The results strongly suggested that the efficient influx of extracellular l-glutamate via the GltT-GltM transporter system into N. meningitidis elicited efficient internalization of N. meningitidis into HBMEC (see Discussion).
Meningococcal internalization was less efficient under l-glutamate-limited conditions.To further examine the possible relationship between l-glutamate influx via the GltT-GltM l-glutamate transporter system and meningococcal invasion, we examined meningococcal infection of HBMEC under l-glutamate-limited conditions in which l-glutamine and FBS were also omitted to eliminate additional l-glutamate (see Materials and Methods). Adherence of both gltT+-gltM+ HT1125 and the ΔgltT-ΔgltM HT1414 N. meningitidis strains was not affected by the l-glutamate concentration in the medium (Fig. 5A). On the other hand, the internalization of HT1125 became less efficient under l-glutamate depletion while the invasive activity of HT1414 was not affected under the same conditions (Fig. 5B). These results suggested that the influx of environmental l-glutamate into N. meningitidis served as a cue for meningococcal internalization into host cells (see Discussion).
Adherence and internalization of N. meningitidis to HBMEC under l-glutamate-limited conditions. Bacterial number is represented as CFU. Internalized bacteria were determined as the gentamicin-resistant bacterial number. Ranges represent the mean CFU of adherent bacteria (A) or internalized bacteria (B) per 104 cells in at least four experiments, and error bars represent the standard errors of the means. Open and filled bars indicate the bacterial number of gltT+-gltM+ (HT1125) and ΔgltT-ΔgltM (HT1414) N. meningitidis strains, respectively.
Ezrin accumulation beneath N. meningitidis strains attached to HBMEC.We next examined changes in the host cell's cytoskeleton rearrangement with meningococcal infection by monitoring the localization of ezrin using indirect immunofluorescence. A broad distribution of ezrin was observed throughout noninfected cells (Fig. 6, upper panels). Double immunofluorescence staining of actin and ezrin and simultaneous observation of bacteria by phase-contrast microscopy revealed that ezrin was condensed at the site of bacterial attachment in HBMEC infected with the gltT+-gltM+ N. meningitidis strain (Fig. 6, middle panels). In contrast, ezrin condensation was not observed in cells infected with the ΔgltT-ΔgltM N. meningitidis mutant (Fig. 6, lower panels). These results indicated that, while the exact mechanism is unknown, l-glutamate influx via the GltT-GltM transporter system into N. meningitidis elicited bacteria-induced reorganization of the host's cell cytoskeleton with meningococcal infection (see Discussion).
Immunofluorescence microscopy showing the accumulation of ezrin beneath the gltT+-gltM+ but not the ΔgltT-ΔgltM N. meningitidis strain. The HBMEC monolayer was infected with the gltT+-gltM+ (middle) and ΔgltT-ΔgltM (lower) N. meningitidis strains. A noninfected HBMEC monolayer is also shown in the upper panels. Bacteria and HBMEC were observed by phase-contrast microscopy (left panels). Actin was stained with Alexa Fluor 568-conjugated phalloidin (red channel), and ezrin was immunostained with anti-ezrin monoclonal antibody and Alexa Fluor 488-conjugated rabbit anti-mouse IgG (green channel).
DISCUSSION
In this study, we reported the isolation of three gltT-gltM mutants that were deficient in their internalization into HBMEC. While the mechanisms of meningococcal infection of human cells remain incompletely understood (6, 56), it is generally considered that meningococcal internalization into human epithelial and peripheral endothelial cells occurs as a consequence of meningococcal adhesion because all of the identified meningococcal invasion factors, such as pili, Opc, Opa, and NadA, also function as adhesive molecules (6, 56). While Opc has only an internalization function in infection of HBMEC by capsulated meningococcal strains (53), this idea is still controversial since it is also reported that the adhesion as well as the internalization in HBMEC was decreased in opc mutants regardless of the capsulated state (34). Considering the present knowledge of meningococcal infection of human cells, it was surprising to find in this study that defects in the gltT-gltM genes were primarily related to meningococcal internalization rather than adhesion to human endothelial and epithelial cells. Isolation of N. meningitidis mutants that were more severely defective in meningococcal invasion than adhesion suggests that meningococcal internalization into human cells is biologically distinguishable from adhesion in N. meningitidis infection. The steps involved in meningococcal infections might be more complicated than speculated to date, and many unidentified virulence factors might be involved in the multiple steps of meningococcal infection of human endothelial and epithelial cells.
Monaco et al. and Colicchio et al. reported that GltT l-glutamate transporter is required for the intracellular growth and survival of meningococcus in HeLa cells (28) and mice (9), respectively. We also confirmed in this study that our ΔgltT-ΔgltM mutation partly affected intracellular bacterial survival in HBMEC; however, the effect was not so strong as to explain why the invasion efficiency of the ΔgltT-ΔgltM mutant is approximately 50 times lower than that of the gltT+-gltM+ strain. In addition, Monaco et al. showed that their gltT mutant was internalized as efficiently as the wild-type strain in HeLa cells (28). These discrepancies might have derived from the difference in experimental conditions, including different human cells and N. meningitidis strains. The authors added a centrifugation step in meningococcal infection of HeLa cells (28). While it is not clear how the centrifugation step affects meningococcal infection, it can be speculated that centrifugation may bypass some infectious steps that might be required for meningococcal infection under our experimental conditions. In addition, we used a nonencapsulated meningococcal strain to monitor efficient N. meningitidis internalization since the capsule inhibits meningococcal invasion into cultured human cells (57, 59). Conversely, the meningococcal capsule plays a pivotal role in intracellular survival (43). We believe that the GltT-GltM l-glutamate transport system has a direct effect on meningococcal internalization into host cells since N. meningitidis internalization was associated with ezrin accumulation in HBMEC, as indicated by double immunofluorescence analyses. Taken together, the results of our study and previous results with HeLa cells (28) and mice (9) suggest that the GltT-GltM l-glutamate transport system has dual roles in meningococcal invasion and intracellular survival.
It is well known that ABC transporters participate in bacterial pathogenesis by supporting the acquisition of nutrients, such as iron, manganese, zinc, amino acids, sugars, and phosphate in the host or the host cells (reviewed in reference 13). On the other hand, some putative ABC transporters, such as glnQ (a glutamine transporter) in group B streptococcus (49), Rv0986 and Rv0987 (putative ABC transporters) in Mycobacterium tuberculosis (33), upgB (a glycerol-3-phosphate ABC transporter homolog) in Brucella abortus (7) and ssu0835 (a putative ABC transporter) in Streptococcus suis (54), have also been shown to contribute to adhesion or invasion of cells rather than the acquisition of nutrients in host cells. In addition, reverse genetic analysis shows that a putative ABC transporter gene, NMB1880, in N. meningitidis is not found in the genome of the nonpathogenic Neisseria lactamica (42). While the detailed mechanism remains to be further uncovered, the results of this study might offer a clue to the close relationship between the influx of materials into pathogens via the respective ABC transporter and bacterial virulence upon bacterial infection of the hosts.
How does extracellular l-glutamate influx via the GltT-GltM transport system affect meningococcal internalization? l-Glutamate is well known as a signal transmitter to activate the Ras/Erk/CREB signaling cascade in eukaryotic neurons (25, 50), but the same mechanism is not found in prokaryotes. While transcription from σ70- and σ38-dependent promoters is, respectively, inhibited and activated by intracellular potassium glutamate in E. coli (15, 23), it would be less likely that imported l-glutamate directly regulates meningococcal gene transcription. It is possible that the l-glutamate influx via the GltT-GltM transporter is linked to the expression of unidentified meningococcal genes for efficient internalization since some ABC transporters in pathogenic bacteria might be coupled to the transcriptional regulation of virulence genes in pathogenic bacteria. For example, the expression of numerous pathogenic genes including SPI-1 in Salmonella enterica serovar Typhimurium is activated by an ArcAB-TolC efflux pump (63), and the transcription of T3SS in Pseudomonas aeruginosa is inhibited by the magnesium transporter MgtE (1). It is also known that the influx of environmental small molecules is linked to bacterial pathogenesis; the internalization of N. gonorrhoeae into human cells is inhibited by the influx of phosphate via the gonococcal major outer membrane porin PorB of serotype A (P.IA) (22, 55), and transcript levels of many virulence genes in Bordetella pertussis are reduced under l-glutamate-depleted conditions (29). The transcription of T3SS in P. aeruginosa is also induced under a low-Ca2+ environment by the elevation of cyclic AMP (cAMP), which induces transcription in cooperation with Vfr, a functional homologue of the E. coli cAMP receptor protein (64). l-Glutamate influx via the GltT-GltM transporter might be coupled to secondary signaling to modulate N. meningitidis pathogenesis for efficient meningococcal internalization. Further analyses of gene expression under the control of gltT-gltM genes and l-glutamate-depleted conditions would help in elucidating the relationship between l-glutamate influx via the GltT-GltM transport system and meningococcal pathogenesis.
Since l-glutamate uptake via the GltT-GltM l-glutamate transporter is less dependent on sodium concentration (28), it can be considered that l-glutamate influx via the GltT-GltM l-glutamate transport system might primarily depend on extracellular l-glutamate concentration. It seems reasonable that l-glutamate influx via the GltT-GltM l-glutamate transport system is linked to meningococcal invasive activity in human cells because l-glutamate concentration is higher in human plasma (approximately 90 μM) (28, 31) than in human CSF (approximately 10 μM) (14, 31). In fact, meningococcal internalization became inefficient under l-glutamate-depleted conditions; however, the inefficiency of N. meningitidis internalization decreased to approximately 1/10 under l-glutamate-depleted conditions while the internalization of the ΔgltT-ΔgltM mutant was approximately 50 times less than internalization of the gltT+-gltM+ strain. This discrepancy can be explained as follows: the gltT+-gltM+ N. meningitidis strain may not mimic the ΔgltT-ΔgltM mutant under l-glutamate-depleted conditions. While the bacteria and HBMEC were washed many times with PBS to remove residual l-glutamate from the medium, we cannot exclude the possibility that remaining l-glutamate or l-glutamine could have affected the meningococcal internalization under l-glutamate-depleted conditions. Alternatively, it is also possible that l-glutamate is excreted or synthesized by HBMEC from the intracellular glutamate/glutamine pool or residual glutamate/glutamine in AM(−S).
The MCE-related domain is found in the deduced amino acid sequence of NMB1964, so that NMB1964 was named gltM in this study. MCE proteins have been characterized as mycobacterial invasion factors (8, 24, 38), and in this study three independent gltT and gltM mutants were isolated as invasion-defective mutants. Furthermore, it was identified in this study that the GltM protein was exposed to the meningococcal surface and that its level was reduced in gltT mutants. From these results, we first speculated that GltM protein alone might contribute to N. meningitidis internalization into human cells; however, this was rejected by the observation that the mutant HT1539, in which a normal amount of GltM was expressed in a gltT-deficient genetic background, could not be efficiently internalized into HBEMC. Moreover, a close relationship was found between meningococcal internalization into HBMEC and l-glutamate influx into N. meningitidis via the GltT-GltM l-glutamate transport system. In addition, meningococcal internalization into HBMEC was not inhibited by adding an excess amount of the purified hydrophilic domain of a GltM recombinant protein during the infection assay (data not shown). From these data, we hypothesize that extracellular l-glutamate influx via the GltT-GltM l-glutamate transporter in N. meningitidis enhanced the efficient internalization of N. meningitidis into human cells. However, since the internalization of N. gonorrhoeae is affected by the structural variation of PorB of serotype B (3, 55), it remains possible that GltM itself functions as an invasion factor if l-glutamate uptake is required to maintain the proper conformation of the GltM protein to facilitate meningococcal internalization.
While gltM is essential for l-glutamate uptake via the GltT-GltM l-glutamate transport system, the function of GltM in l-glutamate uptake is still unclear. Since the purified hydrophilic domain of GltM did not bind to l-[3,4-3H]glutamic acid (data not shown), it is unlikely that GltM functions as an extracellular substrate-binding protein to collect environmental l-glutamate. In E. coli, extracellular maltose is imported into the periplasmic space through the outer membrane LamB porin (also known as λ phage receptor) and further transported into the cytoplasmic space via the maltose ABC transporter complex MalFGK2 (reviewed in reference 11). Since computer analyses by PSIPRED (26) suggests that β-sheet structures are regularly found in the deduced amino acid sequences of GltM (data not shown), the GltM protein might form a “glutamate porin” like E. coli LamB at the outer membrane in N. meningitidis as a component of GltT-GltM l-glutamate transport system.
It was also unexpectedly found that the GltM protein was degraded in gltT mutants. While we could not detect the GltT protein by Western blotting in this study, GltT is speculated to be a minor protein in N. meningitidis. With limited GltT protein, the amount of GltM and GltT proteins might be tightly regulated at the posttranslational level in N. meningitidis to maintain the optimal function of the GltT-GltM l-glutamate transport system. This might also reflect the importance of the GltT-GltM l-glutamate transport system for nutrient acquisition as well as meningococcal pathogenesis. Further analyses of the exact mechanism of GltM degradation might elucidate meningococcal pathogenesis as well as the protein degradation mechanism in N. meningitidis.
Finally, it should be emphasized again that N. meningitidis internalization is experimentally distinguishable from adhesion in N. meningitidis infection to human endothelial and epithelial cells. Since the identified meningococcal invasion factors generally function as adhesin molecules, our finding suggests that there remain many unresolved mechanisms involved in meningococcal internalization into human endothelial and epithelial cells. Comprehensive genome analyses such as mining of the N. meningitidis genome by reverse genetics and STM screening to isolate infection-defective mutants will promote understanding of the unresolved N. meningitidis pathogenesis.
ACKNOWLEDGMENTS
We thank Makoto Ohnishi and Tatsuo Yanagisawa for their helpful discussions.
This work was supported by grants from the Ministry of Health, Welfare and Labor of Japan. H.T. was also supported by a grant from the Ministry of Education, Science and Culture of Japan (grant numbers 16790265 and 21790435).
FOOTNOTES
- Received 11 May 2010.
- Returned for modification 17 June 2010.
- Accepted 8 October 2010.
- Accepted manuscript posted online 18 October 2010.
↵† Supplemental material for this article may be found at http://dx.doi.org/10.1128/IAI.00497-10.
- Copyright © 2011, American Society for Microbiology
