NlpI Contributes to Escherichia coli K1 Strain RS218 Interaction with Human Brain Microvascular Endothelial Cells

Abstract

Escherichia coli K1 is the most common Gram-negative bacillary organism causing neonatal meningitis. E. coli K1 binding to and invasion of human brain microvascular endothelial cells (HBMECs) is a prerequisite for its traversal of the blood-brain barrier (BBB) and penetration into the brain. In the present study, we identified NlpI as a novel bacterial determinant contributing to E. coli K1 interaction with HBMECs. The deletion of nlpI did not affect the expression of the known bacterial determinants involved in E. coli K1-HBMEC interaction, such as type 1 fimbriae, flagella, and OmpA, and the contribution of NlpI to HBMECs binding and invasion was independent of those bacterial determinants. Previous reports have shown that the nlpI mutant of E. coli K-12 exhibits growth defect at 42°C at low osmolarity, and its thermosensitive phenotype can be suppressed by a mutation on the spr gene. The nlpI mutant of strain RS218 exhibited similar thermosensitive phenotype, but additional spr mutation did not restore the ability of the nlpI mutant to interact with HBMECs. These findings suggest the decreased ability of the nlpI mutant to interact with HBMECs is not associated with the thermosensitive phenotype. NlpI was determined as an outer membrane-anchored protein in E. coli, and the nlpI mutant was defective in cytosolic phospholipase A₂α (cPLA₂α) phosphorylation compared to the parent strain. These findings illustrate the first demonstration of NlpI's contribution to E. coli K1 binding to and invasion of HBMECs, and its contribution is likely to involve cPLA₂α.

Neonatal Gram-negative bacillary meningitis continues to be an important cause of mortality and morbidity (12, 26, 35). The key aspect of pathogens associated with neonatal bacterial meningitis is related to their ability to traverse the blood-brain barrier (BBB), but the microbe-host interactions involved in microbial traversal of the BBB remain incompletely understood (17). Escherichia coli K1 is the most common Gram-negative bacillary organism causing neonatal meningitis (12, 26, 35), and most cases of neonatal E. coli meningitis develop via hematogenous spread, but it is incompletely understood how circulating E. coli K1 traverses the BBB (16).

The BBB is a structural and functional barrier formed by brain microvascular endothelial cells to protect the brain from microbes and toxins circulating in the blood. The in vitro BBB model has been developed by isolation and cultivation of human brain microvascular endothelial cells (HBMECs) (31). This in vitro model makes it feasible to study the mechanisms on how meningitis-causing pathogens cross the BBB. It has been shown that E. coli K1 binding to and invasion of HBMECs is a prerequisite for its penetration into the brain (3, 13, 14, 17, 18), but the microbe-host interactions involved in HBMEC binding and invasion remain incompletely understood.

NlpI is a lipoprotein that was first identified in E. coli K-12 (22). The nlpI mutant of E. coli K-12 exhibited growth defect at 42°C on low-salt medium. This phenotype can be suppressed by a mutation on the spr gene. The spr single mutation of E. coli K-12 also exhibited the similar thermosensitive phenotype (32). In addition, NlpI is shown to be an important component of Crohn's disease-associated E. coli strain LF82 (O83:H1) to interact with intestine epithelial cells (5, 6). Deletion of nlpI in E. coli strain LF82 decreases expression of type 1 fimbriae and flagella (5). However, it is unknown whether NlpI is involved in extraintestinal E. coli infection such as meningitis.

While investigating the microbe-host interactions involved in E. coli K1 binding to and invasion of HBMECs, we noted that deletion of nlpI decreased the ability of E. coli K1 to interact with HBMECs. The present study examined and characterized the role of NlpI in E. coli K1 binding to and invasion of HBMECs.

MATERIALS AND METHODS

Bacterial strains and culture condition.

E. coli K1 strain RS218 (O18:K1:H7) is the cerebrospinal fluid isolate from a neonate with meningitis (2, 29). In the present study the spontaneous streptomycin-resistant derivative of RS218 and its derivatives were used. The streptomycin-resistant strain was designated as the wild-type RS218 (Table 1) . The locked-ON and locked-OFF mutants of strain RS218, constructed as previously described (33), have their fim promoter (the promoter control the expression of type 1 fimbria) fixed in the “ON” and “OFF” orientations, respectively. The locked-ON mutant, therefore, constitutively expresses type 1 fimbriae, whereas the locked-OFF mutant dose not express type 1 fimbriae (33). E. coli strains were grown at 37°C overnight in Luria broth (LB) statically unless otherwise specified.

TABLE 1.

Strains and plasmids used in this study

Strain or plasmid	Relevant information	Source or reference
Strains
RS218	E. coli K1RS218 isolated from the cerebrospinal fluid of a neonate with meningitis	2, 29
nlpI mutant	RS218 mutant with an nlpI deletion	This study
spr mutant	RS218 mutant with an spr deletion	This study
nlpI spr mutant	RS218 mutant with an nlpI spr double deletion	This study
ompA mutant	RS218 mutant with an ompA deletion	34
Locked-ON mutant	RS218 mutant with the fim promoter fixed in the on orientation	33, 37
Locked-OFF mutant	RS218 mutant with the fim promoter fixed in the off orientation	33, 37
CHT062	The locked-OFF mutant with an ompA deletion	34
CHT118	The locked-OFF mutant with an nlpI deletion	This study
CHT135	RS218 with an nlpI ompA double deletion	This study
CHT136	The locked-OFF mutant with an nlpI ompA double deletion	This study
Plasmids
pCL1920	Low-copy-number plasmid	20, 32
pNI1	pCL1920 harboring the nlpI gene and its promoter region	32
pNI10	pTrc99A harboring the nlpI gene, which is under the control of the promoter on the plasmid	32
pN10-His	Derivative of pNI10 encoding C-terminal His6-tagged NlpI	This study

Construction of mutants.

Mutants with deletion of nlpI and/or spr were constructed by the PCR-based method described by Datsenko and Wanner (10), using the primers NK-nlpI-F and NK-nlpI-R for nlpI deletion and the primers NK-spr-F and NK-spr-R for spr deletion (Table 2). Briefly, the loci of the nlpI and spr genes in the mutants were replaced with chloramphenicol cassettes. To construct the nlpI spr double mutant, the chloramphenicol cassette in the original nlpI locus of the nlpI single mutant was removed as previously described (10), and then the spr mutation was introduced by replacing the spr gene with the chloramphenicol cassette. The mutants derived from RS218, CHT118 (the locked-OFF mutant with nlpI deletion), CHT135 (the OmpA mutant with nlpI deletion), and CHT136 (the locked-OFF and ompA mutant with nlpI deletion) were constructed by introducing the nlpI deletion mutation into the locked-OFF, ompA, and ompA locked-OFF mutants, respectively, which were derived from our previous studies (33, 34) (Table 1).

TABLE 2.

Primers used in this study

Primer	Sequence (5′-3′)
NK-nlpI-F	ATGAAGCCTTTTTTGCGCTGGTGTTTCGTTGCGACAGCACTTACGCTTGC CATATGAATATCCTCCTTAG
NK-nlpI-R	CTATTGCTGGTCCGATTCTGCCAGGTCATCTTGGTCCTGGCCCAGGAGCGGTGTAGGCTGGAGCTGCTTC
NK-spr-F	ATAACGATATTTGTCGTTAAGGACTTCAAGGGAAAACAAACAACATGGTCCATATGAATATCCTCCTTAG
NK-spr-R	AGCTGCGGCTGAGAACCCGGCGTGCTTCGTTGTAACGCTTCTTCCAGTATGTGTAGGCTGGAGCTGCTTC
NI10-F	GTGGATCCCTAATGATGATGATGATGATGTTGCTGGTCCGATTCTGCCA
NI10-R	GAGGATCCGCTGTTTTGGCGGATGAGAGAAG
Ch1-F	AGTAATGCTGCTCGTTTTGC
Ch1-R	GACAGAGCCGACAGAACAAC

Complementation of the RS218 nlpI mutant.

Complementation experiments were performed with pCL1920 (vector control) and pNI1, which is a pCL1920-derived plasmid harboring the nlpI gene and its promoter region (32) (Table 1). Both plasmids were kindly provided by Akiko Okuda-Tadokoro (Osaka Ohtani University, Osaka, Japan). pNI1 was used to transform the nlpI mutants, while pCL1920 used to transform the wild-type E. coli K1 strain RS218, as well as the nlpI mutant to serve as the vector control.

Invertible element orientation assay.

The invertible element orientation assay utilized the asymmetrical digestion site of SnaBI within the invertible element as previously described (21). A 602-bp fragment containing the invertible element was amplified from both the phase-ON and the phase-OFF bacteria using the primers Ch1F and Ch1R (Table 2). When the invertible element is in the ON position, SnaBI digestion will produce fragments of 404 and 198 bp, whereas the restriction enzyme digestion will make fragments of 442 and 160 bp for the OFF position. The SnaBI-digested PCR products were separated on a 6% acrylamide gel. To quantify the percentage of phase-ON bacteria, a standard curve was prepared as described previously (3) except that defined numbers of the type 1 fimbria locked-ON and locked-OFF bacteria were used as PCR templates. The intensities of each band of the DNA fragments were determined by using the ImageJ program downloaded from the National Institutes of Health website (http://rsb.info.nih.gov).

Yeast aggregation assay.

Yeast aggregation titers of E. coli strains were measured as described previously (30). Briefly, E. coli cultures at an optical density at 530 nm (OD₅₃₀) of 0.4 were subjected to serial 2-fold dilutions in phosphate-buffered saline (PBS) and mixed with commercial baker's yeast cells (5 mg/ml). Aggregation was monitored visually, and the titer was recorded as the highest dilution giving a positive aggregation result.

Construction of the plasmid (pNI10-His) harboring the gene encoding C-terminal His₆-tagged NlpI.

pNI10 is a pTrc99A-derived plasmid harboring the nlpI gene (Table 1) (32), which was provided by Akiko Okuda-Tadokoro. To add the sequence encoding His₆ to the 3′ end of nlpI in this plasmid, pNI10 was served as the template for PCR using the primers NI10-F and NI10-R (Table 2). From the 5′ end to the 3′ end, NI10-R contained a BamHI site and the reverse complement sequence of His₆ fused with the reverse complement sequences of the 3′ end of nlpI. NI10-F contained a sequence downstream of nlpI in pNI10 with a BamHI site at its 5′ end. The PCR product was digested with BamHI and self-ligated to form the new plasmid, pNI10-His encoding C-terminal His₆-tagged NlpI.

E. coli cell fractionation and localization of NlpI in E. coli.

E. coli cells harboring pNI10-His were ruptured by French press at 8,000 lb/in². The unbroken cells in the resulting lysate were removed by centrifugation at 12,000 × g for 4 min at 4°C. The supernatant was then centrifuged at 100,000 × g for 1 h at 4°C. The resulting supernatant contained the bacterial soluble proteins, while the resulting pellet contained the bacterial crude membrane fraction.

Separation of inner and outer membrane from the crude membrane fraction was achieved by sucrose density gradients according to the methods described by Schnaitman (25) with a minor modification. Briefly, the crude membrane fraction was layered on discontinuous sucrose gradient (4 ml of 70%, 4 ml of 54%, and 4 ml of 20% sucrose) in a 12.5-ml centrifuge tube. The sample was centrifuged at 260,000 × g for 10 h at 4°C. After centrifugation, two bands were seen in the centrifuge tubes. The upper band contained the inner membrane-enriched sample, while the lower band contained the outer membrane-enriched sample (25). The gradient was fractionated by collecting 1-ml sample from the top of the centrifuge tubes each time. The existence of NlpI and OmpA in the fractions was determined by Western blot analysis with the anti-His₆ antibody and OmpA antiserum. The expression of OmpA was used to represent the outer membrane fractions.

Western blot analysis of OmpA and flagellin.

To measure the level of OmpA expression in E. coli membranes, the crude membrane fractions were prepared as described above. Equal amounts of membrane fractions from the wild-type RS218 and the nlpI mutant were separated by sodium dodecyl sulfate (SDS)-polyacrylamide gel electrophoresis and transferred to nitrocellulose membranes. The rabbit antiserum against OmpA (1:1,000 dilution) and the goat alkaline phosphatase-conjugated anti-rabbit IgG antibodies (1:3,000 dilution) were used as the primary and secondary antibodies to detect the protein.

To measure the expression levels of FliC in the E. coli strains, equal amounts of whole-bacterial-cell lysates from the wild-type RS218 and the nlpI mutant were subjected to Western blot analysis similar to the procedures to detect OmpA. The rabbit polyclonal antibody against flagellin (anti-H7) was used as the primary antibody (1:1,000 dilution; Becton Dickinson, Spark, MD), and the goat alkaline phosphatase-conjugated anti-rabbit IgG antibodies (1:5,000 dilution) were used as the secondary antibodies to detect the protein.

Motility assays.

To assess the swimming motility of the nlpI mutant and wild-type RS218, the bacteria were initially grown on LB plates of 1.5% agar. After overnight culture, isolated bacterial colonies were stab inoculated onto LB plates of 0.3% agar and incubated at 37°C for 10 h as described previously (23).

E. coli association and invasion assays in HBMECs.

HBMECs were isolated and cultured as previously described (31). HBMEC cultures were grown in RPMI 1640 containing 10% heat-inactivated fetal bovine serum, 10% Nu-Serum, 2 mM glutamine, 1 mM pyruvate, penicillin (100 U/ml), streptomycin (100 μg/ml), essential amino acids, and vitamins. Invasion assays were performed by using gentamicin protection assay as previously described (33). Briefly, confluent monolayers of HBMECs grown in 24-well plate were infected with 10⁷ bacteria at a multiplicity of infection of 100. After 90 min of incubation at 37°C, the monolayers were washed with PBS, incubated with medium containing gentamicin (100 μg/ml) for 1 h to kill extracellular bacteria, and then washed three times with PBS. HBMECs were lysed by incubation with sterile water at room temperature for 30 min. The released intracellular bacteria were enumerated by culturing on sheep blood agar plates. The invasion frequencies were calculated by dividing the number of internalized bacteria by the number of the original inoculum. The results were presented as relative invasiveness, the percent invasion compared to the invasion frequency of the wild-type RS218, which was arbitrarily set at 100%. Association assays were performed as described for the invasion assay above except that the gentamicin treatment step was omitted.

Analysis of cytosolic phospholipase A₂α (cPLA₂α) in HBMECs.

Monolayers of HBMECs cultured in 100-mm dishes were incubated with the wild-type strain RS218 and the nlpI mutant for different time periods. The cells were lysed in 50 mM Tris buffer (pH 7.4) containing 1% Triton X-100, 150 mM sodium chloride, sodium vanadate, and a protease inhibitor cocktail. Lysates were subjected to SDS-PAGE, followed by a transfer of the proteins to nitrocellulose membranes. The membranes were then analyzed for either phospho-cPLA2α or cPLA2α using specific antibodies (Cell Signal Technologies, Danvers, MA), followed by horseradish peroxidase-tagged conjugate. Bound conjugate was analyzed by luminescence using ECL (Amersham). Bands obtained by Western blotting were quantified and graphically represented by using the ImageJ program.

Statistic analysis.

One-way analysis of variance (ANOVA) test was used for statistical analysis. A P value of ≤0.05 was arbitrarily set as the threshold for statistical significance.

RESULTS

The nlpI mutant of E. coli K1 strain RS218 exhibited defects in binding to and invasion of HBMECs.

To determine whether nlpI plays a role in E. coli K1 binding to and invasion of HBMEC, an nlpI deletion mutant was generated from E. coli K1 strain RS218. The growth characteristics of the nlpI mutant and the wild-type RS218 were similar in LB broth and in the experimental medium used for HBMEC association and invasion assays (data not shown). The RS218 nlpI mutant was then investigated for its ability to bind and invade HBMECs. The nlpI mutant exhibited significantly less binding to HBMECs in comparison with the parent strain (7% ± 2% of the wild type). The mutant also showed only 9% ± 1% of invasion frequency of the wild-type strain (Fig. 1A). In addition to confirm that the phenotypes of the mutant in binding to and invasion of HBMECs were not caused by the polar effects of the nlpI gene deletion, we performed complementation experiments. pNI1 is a pCL1920-derived plasmid harboring the nlpI gene and its promoter region (Table 1) (32) and was used to transform the nlpI mutant. We also used pCL1920 to transform the wild-type RS218 and the nlpI mutant as the vector control. The transformants mentioned above were then subjected to HBMEC association and invasion assays. The association and invasion frequencies of the mutant harboring pNI1 were increased to 63% ± 7% and 67% ± 9% of those of the wild-type RS218 harboring pCL1920, respectively, whereas the association and invasion frequencies of the mutant harboring the empty vector were significantly lower than those of the parent strain harboring the empty vector (Fig. 1B). These results indicated that nlpI is an important determinant for E. coli K1 strain RS218 to adhere to and invade HBMECs.

FIG. 1.

The nlpI deletion decreased the ability of E. coli K1 strain RS218 to interact with HBMECs. (A) The RS218-ΔnlpI isogenic mutant showed significantly decreased HBMEC association and invasion frequencies in comparison with the wild-type RS218. (B) The defects of the mutant were restored by complementation with the nlpI gene. pNI1, which was derived from the vector plasmid, pCL1920, harbors the nlpI gene. Hatched and filled bars indicate relative association and invasion frequencies, respectively, compared to the wild-type strain RS218, which was set at 100%. The association frequencies of the wild-type RS218 (A) and the pCL1920-harboring wild-type RS218 (B) were 25% ± 2% and 19% ± 3% of the original inoculum, respectively, while their invasion frequencies were 0.97% ± 0.13% (A) and 1.09% ± 0.06% (B) of the original inoculum. The data shown are representative of three independent experiments done in triplicate. The results were shown as means ± the standard deviations (SD). *, P values of ≤0.01 between the nlpI mutant and the wild-type strain as determined by ANOVA. +, P values of ≤0.01 between the nlpI mutant strains harboring pNl1 and pCL1920 as determined by ANOVA.

Deletion of nlpI in E. coli K1 strain RS218 did not affect the expression of type 1 fimbriae, flagella, and OmpA.

Type 1 fimbriae, flagella, and OmpA have been shown to be the important bacterial determinants for E. coli K1 strain RS218 to bind to and invade HBMECs (23, 33, 34). In addition, deletion of nlpI in E. coli strain L82 causes significantly decreased expression of type 1 fimbria and flagellum (5). To investigate whether deletion of nlpI decreases the expression of the three bacterial determinants in E. coli K1 strain RS218, we compared their expression and function in the RS218 nlpI mutant with those in the wild-type strain.

Expression of type 1 fimbriae is dependent on a phase variation in which each individual bacterium can alternate between fimbriated and nonfimbriated states, also called as phase-ON and phase-OFF, respectively (1). The phase switching is determined by the orientation of a 314-bp chromosomal region that contains the promoter of fim structure genes and is located upstream of the fimA gene encoding the major subunit of type 1 fimbriae. In E. coli strain LF82, nlpI deletion decreases type 1 fimbria expression through preferentially driving the fim promoter toward the phase-OFF orientation. Thus, we first determined the percentages of E. coli cells with the fim promoter in the phase-ON orientation in the wild-type RS218 and its nlpI mutant cultures by using invertible element orientation assays (33). The percentages of the phase-ON cells in the RS218 wild-type and mutant culture were not significantly different (Fig. 2A). We next examined the function of type 1 fimbriae on the wild-type strain and the mutant by yeast aggregation assay, since type 1 fimbriated bacteria are capable of aggregating yeast cells through their ability to bind mannosylated glycoproteins present on the surface of yeast cells. The yeast aggregation titer of the nlpI mutant was similar to that of the wild-type RS218 (Table 3). These findings demonstrate that the levels of type 1 fimbria expression and function are similar between the wild-type RS218 and the nlpI mutant.

FIG. 2.

NlpI deletion in E. coli K1 strain RS218 did not affect the expression of type 1 fimbriae, flagella, and OmpA. (A) The invertible element orientations assay revealed that the percentages of the cells with the fim promoter in the phase-ON orientation in the wild-type RS218 and the nlpI mutant were similar. The 198- and 160-bp DNA fragments represented the SnaBI digested phase-ON and phase-OFF products (see Materials and Methods). (B) Western blot analysis of FliC levels in the wild-type strain and the RS218 nlpI mutant, using anti-flagellin H7 antibodies. The expression of FliC was similar between the two strains. (C) The swimming motilities of the wild-type strain and the RS218 nlpI mutant on the 0.3% agar plates were similar. (D) Western blot analysis of OmpA in the membrane fraction of the wild-type RS218 and the nlpI mutant. The OmpA expression in the membrane fraction of the two strains was similar.

TABLE 3.

Comparison of yeast aggregation among RS218 and its mutants

Straina	Yeast aggregation titerb:
	Without mannose	With 100 mM mannose
Wild-type RS218	8	−
nlpI mutant	8	−
Locked-ON mutant	16	−
Locked-OFF mutant	−	−

^aAll strains were grown in static LB broth.

^bAgglutination titers were recorded as the highest 2-fold dilution giving a positive aggregation, starting at an OD₅₃₀ of 0.4. −, No aggregation was observed.

The levels of flagellum expression in the RS218 wild-type and nlpI mutant strains were determined by Western blot assay using polyclonal antibodies against flagellin H7. The expression of FliC in the two bacteria did not show significant difference (Fig. 2B). Consistently, the swimming motility of the wild type and the mutant showed no significant difference (Fig. 2C). The results demonstrate that deletion of nlpI in E. coli K1 strain RS218 did not affect the expression and function of flagella.

The total level of OmpA in the membrane fraction in the nlpI mutant and the wild-type strain were measured by Western blot assays with anti-OmpA rabbit serum. The two E. coli strains exhibited similar levels of OmpA in their bacterial membrane fractions (Fig. 2D).

Roles of type 1 fimbriae, OmpA, and NlpI in E. coli K1-HBMEC interaction.

Our results thus far indicated that NlpI is the additional determinant for E. coli K1 strain RS218 to bind to and invade HBMECs. We next examined the roles of NlpI, OmpA, and type 1 fimbriae in E. coli K1 strain RS218's binding to and invasion of HBMECs by constructing mutants deleted of ompA and/or nlpI in the locked-OFF background and examining their ability to bind to and invade HBMECs.

As shown in Table 4, single gene deletion mutants of ompA and nlpI, as well as the type 1 fimbria locked-OFF mutant, exhibited significantly decreased HBMEC binding and invasion compared to the parent strain. Of interest, the double gene deletion mutant of ompA and nlpI (designated as CHT135), as well as the ompA or nlpI deletion mutant in the type 1 fimbria locked-OFF background (designated as CHT062 and CHT118, respectively), exhibited significantly decreased HBMEC binding and invasion compared to the single gene deletion mutant or the locked-OFF mutant. These findings suggest that OmpA, NlpI, and type 1 fimbriae are likely to contribute to HBMEC binding and invasion independent of each other. This concept was further substantiated by the demonstration that the mutant deleted of ompA and nlpI in the background of type 1 fimbria locked-OFF mutant (designated as CHT136) was least able to bind to and invade HBMEC (Table 4).

TABLE 4.

Comparison of HBMEC binding and invasion frequencies of E. coli K1 strain RS218 versus its mutants

Strain	Mean frequency (%) ± SD
	Association	Invasion
Wild-type RS218	100 ± 6.67	100 ± 8.48
ompA mutant	52.55 ± 10.51	40.79 ± 5.32
nlpI mutant	8.61 ± 2.07	9.77 ± 1.03
Locked-OFF mutant	30.05 ± 3.26	8.44 ± 1.74
CHT062 (ΔompA/locked-OFF)	3.30 ± 0.13	0.75 ± 0.10
CHT118 (ΔnlpI/locked-OFF)	0.06 ± 0.02	0.10 ± 0.02
CHT135 (ΔnlpI/ΔompA)	0.13 ± 0.07	0.09 ± 0.07
CHT136 (ΔnlpI/ΔompA/locked-OFF)	<0.01	<0.01

Deletion of spr did not restore RS218 nlpI mutant's ability to bind to and invade HBMECs.

In E. coli K-12 strains, nlpI mutants exhibited growth defect on low-salt medium at 42°C, and such a phenotype is suppressed by mutation of the spr gene. Single gene mutation of spr also exhibits a similar thermosensitive phenotype in E. coli K-12 strain (32). The single deletion mutants of nlpI and spr, as well as the nlpI-spr double mutant of E. coli K1 strain RS218, were also found to be similar to the E. coli K-12 mutants in terms of their thermosensitive phenotypes (data not shown). We next investigated whether the nlpI deletion-caused thermosensitive phenotype in E. coli K1 strain RS218 is associated with the decreased abilities to interact with HBMECs. We examined and compared the nlpI and spr single mutants and the nlpI spr double mutant along with the wild-type RS218 for their abilities to bind to and invade HBMECs. The nlpI, spr, and nlpI spr mutants exhibited 9% ± 1%, 101% ± 17%, and 0.3% ± 0.07% of the association frequencies of the wild-type strain, respectively, whereas the same mutants showed 11% ± 1%, 79% ± 7%, and 6% ± 2% of the invasion frequencies of the wild-type strain, respectively (Fig. 3). Unlike the phenotype of thermosensitivity, deletion of spr in the nlpI mutant (i.e., the spr nlpI double mutant) did not restore the abilities of the nlpI mutant to interact with HBMECs. In addition, the spr mutant did not exhibit significantly decreased ability to interact with HBMECs compared to the wild-type strain. These findings suggest that the thermosensitive phenotype of the nlpI mutant was not correlated with its decreased association with HBMEC.

FIG. 3.

Effects of spr deletion on the RS218-ΔnlpI mutant and the wild-type strain in terms of their abilities to interact with HBMECs. Hatched and filled bars indicate relative association and invasion frequencies, respectively, compared to the wild-type strain RS218, which was set at 100%. The association and invasion frequencies of the wild-type RS218 were 21% ± 3% and 0.75% ± 0.06%, respectively. The data shown are representative of three independent experiments done in triplicate. The results were shown as means ± the SD. *, P ≤ 0.01 between the nlpI or spr nlpI mutants versus the wild-type strain or spr mutant as determined by ANOVA.

Localization of NlpI in E. coli.

Since NlpI is found to be involved in E. coli interaction with host cells, we next investigated the localization of this protein in E. coli. The plasmid pNI10-His harboring nlpI fused with His₆ tag sequence at its 3′ end was constructed and transferred into the E. coli strain DH5α. The transformant was able to express C-terminal His₆-tagged NlpI, and the existence of the recombinant NlpI was detected by Western blot analysis with the anti-His₆ antibody. Similar to OmpA (a known E. coli outer membrane protein), NlpI was mainly located in bacterial crude membrane (Fig. 4A), suggesting that NlpI is a membrane protein.

FIG. 4.

Localization of NlpI. (A) NlpI was mainly located in crude membrane fraction of E. coli. Soluble, the soluble fraction of French press-ruptured bacterial cell; Membrane, the crude membrane fraction of French press-ruptured bacterial cells. (B) NlpI was mainly located in the outer membrane fraction. The crude membrane of ruptured E. coli cells was subjected to sucrose density gradient analysis to separate the inner and outer membrane fractions. Fractions corresponding to the inner membrane (IM), the intermediate density material (M), and the outer membrane (OM) were shown. The existence of NlpI and OmpA was determined by Western blot analysis with the anti-His₆ antibody and the OmpA antiserum, respectively.

To determine whether NlpI is an outer membrane protein, the bacterial crude membrane was subjected to sucrose density gradients analysis. Similar to OmpA, NlpI was mainly located in the outer membrane-enriched fractions, suggesting that NlpI is an outer membrane protein (Fig. 4B).

NlpI contributes to cPLA₂α activation in HBMECs.

We have previously shown that E. coli K1 strain RS218 triggers a cascade of host signaling events in HBMEC, facilitating its entry into host cells, which include cPLA₂α (15-17). This was shown by our demonstration that RS218 invasion of HBMEC was significantly inhibited by cPLA₂α inhibitor and RS218 invasion was significantly less in brain endothelial cells derived from cPLA₂α^−/− mice (8). However, the determinants of RS218 trigger cPLA₂α activation are not known.

The HBMECs were incubated with the wild-type strain RS218 and the nlpI mutant, and lysates were examined for phospho-cPLA₂α and cPLA₂α by using specific antibodies. The ratios of phospho-cPLA₂α to cPLA₂α were calculated and are represented as the fold change in comparison to wild-type RS218. Western blot analysis revealed that cPLA₂α phosphorylation in response to the wild-type RS218 occurred in a time-dependent manner in HBMECs, whereas its phosphorylation in response to the nlpI mutant was minimal (Fig. 5). These findings suggest that NlpI contributes to E. coli K1 invasion of HBMECs, most likely involving cPLA₂α.

FIG. 5.

Comparative analysis of host cell cPLA₂α activation by the wild-type strain RS218 and its nlpI mutant. (A) Confluent monolayers of HBMECs were incubated with RS218 or its nlpI mutant for various time points. Cell lysates were then analyzed for phospho-cPLA₂α (p-cPLA₂α) or cPLA₂α by using specific antibodies in Western blot analysis. (B) p-cPLA₂α and cPLA₂α band intensities were estimated densitometrically and represented as the fold increase of p-cPLA₂α to cPLA₂α.

DISCUSSION

E. coli meningitis develops as a consequence of its penetration across the BBB. E. coli penetration of the BBB is a multistep process, including E. coli binding to and invasion of HBMECs, the major component cells of the BBB. In the present study, we identified and characterized a novel bacterial factor, NlpI, contributing to E. coli K1 binding to and invasion of the BBB by using HBMECs culture as the in vitro BBB model.

The RS218-nlpI mutant showed significantly lower binding and invasion frequencies in HBMECs compared to the wild-type RS218, indicating that NlpI is a bacterial determinant involved with E. coli K1 strain RS218 for its binding to and invasion of HBMECs. The binding and invasion abilities of the RS218 nlpI mutant complemented with the nlpI gene in the plasmid pNI1 were not fully restored, exhibiting ca. 63 and 67% of the binding and invasion frequencies of wild-type RS218 containing the low-copy-number empty vector, pCL1920. Although pN1 is a low-copy-number plasmid, the copy numbers of the nlpI gene in the nlpI mutant harboring pN1 are still higher than that in the wild-type RS218 (only one copy in the chromosome in the wild-type strain). We speculate that this incomplete complementation might be due to a higher amount of NlpI being provided by the complemented plasmid. This speculation is suggested by the demonstration that the ability to bind to intestinal epithelial cells of the LF82 nlpI mutant is not significantly restored by complementation with nlpI in a multicopy recombinant plasmid (5).

Although in E. coli strain LF82 deletion of nlpI decreases the expression of type 1 fimbriae and flagella (5), we demonstrated that in E. coli K1 strain RS218 deletion of nlpI dose not affect the expression of the two bacterial factors. These findings suggest that (i) the decreased abilities of nlpI mutant of E. coli K1 strain RS218 to interact with HBMECs is not through affecting expression level of type 1 fimbriae and flagella and (i) the cross-talks of NlpI with type 1 fimbriae and flagella are different between E. coli K1 strain RS218 and E. coli strain LF82. Similarly, deletion of flagella has been shown to decrease type 1 fimbria expression in E. coli strain LF82, but it does not affect type 1 fimbria expression in E. coli K1 strain RS218 (4, 23). All of these differences between strain RS218 and strain LF82 might be due to their different genomic backgrounds. It is likely that the expression of nlpI may affect the expression of type 1 fimbria and flagella through pathways whose components only encoded in LF82 and not in RS218. This hypothesis is supported by finding that type 1 fimbria expression can be regulated by E. coli factors that exist in some but not all E. coli strains, which include HbiF (or IpbA) in E. coli K1 strain RS218 and the uropathogenic E. coli strain CFT073, as well as IpuA and IpuB in CFT073 (7, 37). Additional studies are needed to elucidate the reasons for these differences between E. coli K1 strain RS218 and E. coli strain LF82.

OmpA is another bacterial factor involved in E. coli K1 interaction with HBMECs (34). Similar to those of type 1 fimbriae and flagella, expression of OmpA was not affected by deletion of nlpI. Although the single mutants of the three bacterial factors exhibited significantly lower abilities interacting with HBMECs compared to the wild-type strain, the double mutants of either two of the factors showed even lower abilities to interact with HBMECs than the single mutants. The triple mutant of the three bacterial factors has the lowest ability to interact with HBMECs. These results suggest that NlpI works with the other two bacterial factors additively or synergistically to interact with HBMECs. This result is consistent with our finding that the effects of nlpI deletion on E. coli K1 strain RS218-HBMEC interaction is not through affecting the expression of the other known bacterial determinants such as type 1 fimbriae, flagella, and OmpA.

nlpI encodes a prolipoprotein of 294 amino acids, including an N-terminal signal sequence of 18 amino acids and a consensus lipobox sequence for lipid modification (36). The deduced residue at the +2 position of the mature lipoprotein is serine (Ser-20). According to the “+2 rule” for lipoprotein sorting in E. coli envelope (11, 27), NlpI is likely to be anchored at the outer membrane but not at the inner membrane. Our finding that C-terminal His₆-tagged NlpI was located in the outer membrane fraction of E. coli confirmed the prediction based on the “+2 rule.” The recombinant NlpI protein, however, did not affect E. coli K1 binding to and invasion of HBMECs (data not shown), and the failure to block E. coli K1 binding may be due to specific lipoprotein characteristics of NlpI (e.g., the NlpI protein may need to be properly located in the bacterial outer membrane so that it can interact with its corresponding receptor on HBMECs). Alternatively, this bacterial determinant may indirectly affect E. coli K1 binding to HBMEC. The deduced amino acid sequence of nlpI contains five tetratricopeptide repeat (TPR) motifs (22, 32, 36). The TRP motifs are known to mediate intermolecular protein-protein interactions, leading to formation of protein complex (9, 19, 24). NlpI may work through the TPR domains to interact with other bacterial factors to form protein complexes that then facilitate the E. coli K1-HBMEC interaction.

We have shown that E. coli binding to and invasion of HBMEC involves microbe-specific host signaling molecules, including host cPLA₂α (15-17). We showed here that serine phosphorylation of cPLA₂α is defective with the nlpI mutant compared to the wild-type strain RS218, suggesting that the defects in HBMEC binding and invasion with the nlpI mutant is likely to stem from its inability to activate cPLA₂α.

In summary, we showed for the first time that NlpI played a role in E. coli K1 binding to and invasion of HBMECs. Unlike in the Crohn's disease-associated E. coli strain LF82, in E. coli K1 strain RS218 deletion of nlpI did not exhibit decreased expression of type 1 fimbria, flagella, and OmpA, which are known bacterial determinants involved in E. coli-HBMEC interaction. The contribution of this bacterial factor to the E. coli K1-HBMEC interaction was found to be independent of that of OmpA, type 1 fimbria, and flagella, which have been shown to be important E. coli K1 factors to interact with HBMECs (15, 18, 28, 33, 34). This is also the first demonstration that NlpI is located in outer membrane of bacterial cells and involved in cPLA₂α activation. Although deletion of spr suppressed the nlpI mutant's phenotype of thermosensitive growth on low-salt medium (32), deletion of spr was not able restore nlpI mutant's ability to interact with HBMECs. Studies are in progress to further characterize how NlpI is involved in the pathogenesis of E. coli meningitis.