DECIPHERING THE ROLES OF OUTER MEMBRANE PROTEIN A EXTRACELLULAR LOOPS IN THE PATHOGENESIS OF ESCHERICHIA COLI K1 MENINGITIS

domains involved in the of brain pathological changes were more severe and pronounced in mice infected with loop 4 mutants. Mutations in loop 1 and loop2 of OmpA rendered the E. coli K1 unable to cause brain damage in mice and therefore appeared similar to control uninfected mice. The results are representative of five independent experiments with 12 animals per group. Scale

Outer membrane protein A (OmpA) has been implicated as an important virulence factor in several Gram-negative bacterial infections such as Escherichia coli K1, a leading cause of neonatal meningitis associated with significant mortality and morbidity. In the present study, we generated E. coli K1 mutants that express OmpA in which three or four amino acids from various extracellular loops were changed to alanines and examined their ability to survive in several immune cells.
We observed that loop regions 1 and 2 play an important role in the survival of E. coli K1 inside neutrophils and dendritic cells and loop regions 1 and 3 are needed for survival in macrophages. Concomitantly, E. coli K1 mutants expressing loop 1 and 2 mutations were unable to cause meningitis in a newborn mouse model. Of note, mutations in loop 4 of OmpA enhance the severity of the pathogenesis by allowing the pathogen to survive better in circulation and to produce high bacteremia levels. These results demonstrate, for the first time, the roles played by different regions of extracellular loops of OmpA of E. coli K1 in the pathogenesis of meningitis and may help in designing effective preventive strategies against this deadly disease.
Escherichia coli K1 is a prominent Gram-negative bacterium that causes meningitis in neonates with case fatality rates ranging from 5 to 30% of infected infants (1)(2)(3)(4). Those who survive are often left with permanent neurological dysfunction such as hearing loss, mental retardation, and cortical blindness (5,6). Despite the use of advanced antibiotics, the morbidity and mortality rates associated with E. coli K1 meningitis remain unchanged over the last few decades (7,8). In addition, due to a recent surge in antibiotic resistant E. coli K1 strains, the mortality rates will further increase significantly (9,10). Therefore, new modes of prevention are warranted for which the understanding disease pathophysiology is clearly necessary. Studies from this lab have demonstrated that E. coli K1 interacts with human brain microvascular endothelial cells (HBMEC) to enter the central nervous system (11,12). The interaction of the bacterium with HBMEC is mediated by outer membrane protein A (OmpA) of E. coli K1 and a glycoprotein receptor, Ecgp96 on HBMEC (13,14). OmpA initially binds to GlcNAc1-4GlcNAc epitopes of Ecgp96, followed by the peptide portion of the receptor (15). Of note, synthetic peptides that represent loops 1 and 2 of OmpA prevented the E. coli K1 invasion of HBMEC (15). OmpA has been shown to be responsible for conferring serum resistance by binding to a complement regulator protein, C4bbinding protein (C4bp) (16,17). Neonates having lower than the threshold levels of C4bp may be at a higher risk to E. coli K1 meningitis as evidenced when the bacterium treated with adult serum, which contained higher amounts of C4bp, could not invade HBMEC compared to newborn serum treatment (18). Nonetheless, neither of the sera prevents the entry of E. coli K1 into macrophages or dendritic cells (DCs) for which OmpA expression is also needed, indicating that various epitopes of OmpA are involved at different stages of pathogenesis of E. coli K1 meningitis (19,20). Therefore, it is our goal in this study to delineate the roles of different domains of OmpA in interacting with various cells in vitro and their role in a well-established newborn mouse model of meningitis (7,(21)(22)(23).
OmpA is a highly conserved molecule among Gram-negative bacteria, which is a 325 amino acid protein, out of which 177 amino acids at the Nterminal portion are shown to cross the membrane eight times in antiparallel β-strands with four relatively large and hydrophilic surface-exposed loops and short periplasmic turns (24)(25)(26)(27). The remaining C-terminal region sits in the periplasmic area of the bacteria. The function of OmpA is thought to contribute to the structural integrity of the outer membrane along with murein lipoprotein and peptidoglycan associated lipoprotein (28)(29)(30)(31). OmpA also serves as a receptor of colicin and several phages, which is required in F-conjugation (32,33). OmpA-like proteins are present in all 17 genera of Gram-negative bacteria and a comparison of five OmpA homologues from five closely related genera indicated that the β-sheet residues of the N-termini are highly conserved while the extracellular loops have relatively large variations (34). Koebnik has shown that the extracellular loops of OmpA can be deleted without affecting the assembly of the β-barrel structure (35,36). However, it is strongly believed that deletion of portions of extracellular loops affects the function of OmpA (31), although the roles of various loops of OmpA in the pathogenesis of E. coli K1 have not been examined to date. We have generated a variety of OmpA mutants by changing three or four amino acids at a time to alanines and demonstrated that some regions of loops 1, 2, and 4 of OmpA are critical for the invasion of E. coli K1 in HBMEC (37). Here, we further analyze the effects of mutations in OmpA mediated interaction with C4bp, various immune cells, and the occurrence of meningitis in a newborn mouse model of meningitis.

EXPERIMENTAL PROCEDURES
Generation of OmpA mutants of E. coli K1−E. coli K1 strain E44 (OmpA + ) is a rifampin-resistant mutant of the clinical isolate RS218 and E98 (OmpA − ) is an ompA-deletion mutant of E44 (13). Various mutations in OmpA were generated as described previously using the plasmid pKE325 (low copy number), which is a pACYC-derivative (pK194) containing entire ompA gene with its own promoter (37). Briefly, primer pairs containing the desired mutations were synthesized and cartridgepurified. PCR for the plasmid mutation was carried out using PfuTurbo DNA polymerase and pKE325 as the template. A faint band was visualized on agarose gel if the amplification succeeded. DNA was then digested with DpnI and transformed into E. coli DH5α followed by selecting for kanamycin resistance. Plasmids were isolated, and the mutation was verified by sequencing. The correct plasmids were introduced into the ΔompA strain E98.The expression of OmpA on each mutant strain of E. coli was determined by flow cytometry. Overnight grown E. coli strains were centrifuged and washed twice with Hanks' balanced salt solution (HBSS) at room temperature. The bacteria were re-suspended in HBSS and adjusted to 1 × 10 9 CFU/ml. The bacterial suspension was incubated with anti-OmpA antibody (1: 10, 000 dilution) at 37°C for 1h, the unbound antibody was removed by centrifugation and the bacteria were washed with PBS three times. The bacterial pellets were then incubated with FITC-conjugated secondary antibody for 30 min at room temperature, washed and re-suspended in PBS. Flow cytometry analysis was carried out on a Becton-Dickinson Flow cytometer using Cell Quest software and ~10,000 events were recorded.
Isolation of outer membrane proteins and Western blotting-Outer membrane proteins from E. coli K1 were isolated as described previously (38). Various strains of E. coli K1 mutants were grown overnight in LB broth containing 100 μg of rifampin or 50 μg of kanamycin per ml. The bacteria were collected by centrifugation and washed three times with TE buffer (20mM Tris, 10mM EDTA, pH 7.4). The pellet was then resuspended in 1 ml of TE buffer and sonicated three times for a minute each time and centrifuged at 6, 000 × g for 10 min at 4 o C. The pellet was discarded and the supernatant was subjected to ultracentrifugation at 45, 000 × g for 45 min at 4 o C. The pellet was re-suspended in 1% sodium lauryl sarcosinate (SLS) in TE buffer and incubated at 37 o C for 2 h to completely digest the inner membrane. The solution was then subjected to ultracentrifugation at 45, 000 × g for 45 min at 4 o C and the pellet was suspended in 20 µl of 1% sodium dodecyl sulfate (SDS) in TE buffer. 10 µg of outer membrane preparations were separated on 10% SDS polyacrylamide gel electrophoresis and subjected to Western blotting with anti-OmpA antibody (38).

Dot blot analysis of S-fimbriae expression-
Bacterial lysates (50 µg) prepared from 10 7 cfu/ml of various E. coli K1 mutants were spotted onto nitrocellulose membrane and allowed to dry for 15 min. The blot was blocked with 5% milk in PBS, then probed with anti-S-fimbriae antibody (1:1000 dilution) and incubated for 90 min at room temperature. The blot was washed 3 times with PBS/Tween-20 (PBST) and incubated with HRP conjugated anti-mouse secondary antibody for 1 h at room temperature. After five washes with PBST, the blot was developed using super signal West Pico chemiluminescent substrate (Pierce technologies, Rockford, IL. (39). The blot was scanned and the densities of the dots were quantified using ImageJ software (NIH). A score of +++ means that the density was similar to that of WT E. coli K1.
Growth pattern of various E. coli K1 strains−The growth pattern of the OmpA + , OmpA − and the loop mutants of E. coli was analyzed using spectrophotometric turbidity method at 600nm (OD 600 ). 1% of overnight cultures of the strains were sub-cultured in 20 ml of LB broth and OD was taken as the zero hour reading. The cultures were then incubated at 37 o C with constant shaking at 220 rpm and 1 ml of the respective cultures were taken at every 1 h and the OD was measured again.
Yeast aggregation assay-To analyze type 1 fimbriae expression, the mutant strains were subjected to yeast aggregation assay as described previously (40). In short, the mutants were cultured overnight and their optical density was adjusted to 0.4 at 530 nm using sterile PBS. 20 μl of each of the cultures was mixed with 20 μl of a 5mg/ml suspension of commercial grade yeast in sterile PBS on a clean glass slide. Scoring for the agglutination was made based on visual observation of agglutination. A score of ++++ represents strong agglutination within 10 seconds of mixing. Subsequent scoring was based on additional 10-second time periods.
Total polysaccharide content determination−Total keto-deoxyoctulosonate (KDO) estimation was performed as previously described (17). Briefly, 10 10 CFU of bacteria were harvested, pelleted, washed once in PBS, and resuspended in 5 ml cold PBS. Twenty microlitres were removed for quantitative estimation of the bacteria and the remaining suspension was sonicated continuously at 50% power for 4 min on ice. Following sonication, 200 μl was mixed with 0.1 ml of the periodate solution (0.2 M in 9 M phosphoric acid). The tubes were agitated and allowed to stand at room temperature for 20 min. Then, 1 ml of sodium arsenite solution (10% in solution of 0.5 M sodium sulfate 0.1 N H 2 SO 4 ) was added and mixed until a yellow-brown color appeared followed by the addition of 3 ml of thiobarbituric acid (0.6% in 0.5 M sodium sulfate). The tubes were shaken and heated vigorously in a boiling water bath for 15 min. 1 ml of this solution, was transferred to another tube, which contained 1 ml of cyclohexanone. The tube was shaken twice and then centrifuged for 3 min at 1000 × g. The clear upper red-colored cyclo-hexanone phase was collected and the absorbance was measured at 549 nm. The molecular extinction coefficient used was 57, 000. The amount of KDO present in a given sample is determined from the following equation: V × OD 549 /57 = 4.3 × OD 549 /57 = 0.075 × OD 549 , where V is the final volume of the test solution.
Serum survival assays−The bacterial suspensions (10 6 CFU/ml) were incubated with freshly prepared serum diluted to 40% in gelatin-veronal buffer (Sigma, St Louis, MO) (17). Aliquots of 10 μl were removed at various times, serially diluted in saline and dilutions were plated on sheep blood agar. The plates were incubated at 37°C overnight, the colonies counted, percent survival calculated and compared with the survival of wild type E. coli K1.
C4bp binding assays−The binding of C4bp to bacteria was examined by flow cytometry (17).
Overnight grown E. coli K1 strains were centrifuged and washed twice with Hanks' balanced salt solution (HBSS) at room temperature. The bacteria were re-suspended in HBSS and adjusted to 1 × 10 9 cells/ml. The bacterial suspension was incubated with 200 nM purified C4bp at 37°C, the unbound C4bp was removed by centrifugation and the bacteria were washed with PBS three times. The bacteria were then incubated with C4bp antibody (1: 1000 dilution) (Calbiochem, San Diego, CA) at room temperature for 1 h followed by three washings with PBS. The bacterial pellets were then incubated with FITC-conjugated secondary antibodies for 30 min at room temperature, washed and re-suspended in PBS. Flow cytometry analysis was carried out on a Becton-Dickinson Flow Cytometer as described above.

Determination of E. coli K1 entry and survival in dendritic cells (DCs), polymorphonuclear leukocytes
(PMNs) and peritoneal macrophages−Monocyte derived myeloid DCs were isolated from the blood of human volunteers as described previously (19). Mouse DCs were obtained by differentiating bone marrow cells isolated from the femurs and tibias of mice (23). PMNs were isolated from the peripheral blood of the human volunteers and mice as described previously (21,41,42). IRB of CHLA approved the blood withdrawal protocol from human volunteers. Macrophages were isolated from the peritoneal cavity of mice as described earlier (22,43). Total bacteria entered and intracellular bacteria in neutrophils, DCs and macrophages were determined as described previously (19,(21)(22)(23). Respective cells were incubated with bacteria at a multiplicity of infection 10 (cell to bacteria ratio, 1: 10) at 37°C for 2, 4 and 6 h. At different post-infection time periods, the co-cultures were centrifuged at a low speed (700 x g); aliquots from the supernatants were diluted and plated on blood agar. The number of bacteria present in the supernatants was subtracted from the total number of bacteria added to co-cultures to obtain the number of bacteria entered the phagocytes. To assess the intracellular bacteria, the co-cultures were centrifuged at a low speed to pellet the cells, gentamicin added at a final concentration of 100 μg ml -1 to the pellet, and the cells were incubated for an additional 60 min at 37°C. The co-cultures were washed three times, the cells lysed with 0.5% Triton X-100, and the released intracellular bacteria were enumerated by plating the dilutions on blood agar.
Newborn mouse model of meningitis−The animal studies were approved by the Institutional animal care and use committee (IACUC) of The Saban Research Institute at CHLA and followed NIH guidelines for the performance of animal experiments. Three-day old mice were infected intranasally with 10 3 CFU of bacteria in 10 μl of saline (21,22). Control mice received pyrogen free saline through the same route. At different times post-infection, blood was collected from the tail or facial vein and plated on antibiotic containing LB agar. The occurrence of meningitis was assessed by culturing 5 μl of cerebrospinal fluid (CSF) obtained by cisternal puncture, without traumatic tap, into LB broth containing antibiotics (21,22). Mice were perfused intracardially with saline to remove blood as well as contaminating intravascular leukocytes. Various tissues were aseptically harvested, homogenized in sterile PBS, and the bacterial counts were determined by plating ten-fold serial dilutions on rifampicin or kanamycin containing LB agar.
Determination of the blood-brain barrier leakage-Newborn mice infected with E. coli K1 were given an intraperitoneal injection of 2% Evan blue dye at 68 h post-infection or 20 h in case of mice infected with loop 4 mutants. After 4 h, the animals were anesthetized with Nembutal and transcardially perfused with PBS until colorless perfusion fluid was obtained from the right atrium. The brains were harvested, weighed, homogenized in PBS and centrifuged to collect the supernatants. Evans blue intensity in the supernatants was determined by a microplate reader at 550 nm. Calculations were based on the external standards dissolved in PBS. The amount of Evans blue was quantified as micrograms per milligram protein (22).
Histopathological analysis−Half of the brain was fixed in 10% buffered formalin, routinely processed, and embedded in paraffin. 4 to 5-μm sections were cut using Leica microtome and stained with hematoxylin and eosin (H&E) (21,22). The slides were read by a neuropathologist in a blinded fashion.
Cytokine determination−The levels of cytokines in the serum and brain tissue homogenates of infected mice were measured using ELISA kits according to the manufacturer's instructions (Biosource, Carlsbad, CA).
Statistical analysis−ANOVA, Fischer test, Chisquare test and paired Student's t-test were applied. A value of p<0.05 was considered significant.

RESULTS
Changing three or four amino acids to alanines in the extracellular loops had no effect on general phenotypic characteristics of E. coli K1 and the interaction with C4bp−In view of dissecting the roles of various domains of OmpA in the pathogenesis of E. coli K1 meningitis, we have generated several strains of E. coli K1 expressing OmpA in which three or four amino acids in various extracellular loops have been changed to alanines and a diagram showing the location of mutations was depicted in Fig. 1A. Computer modeling studies of mutant OmpA in comparison to wild type OmpA showed no significant differences in the structure (37). The plasmids containing mutations were transformed into E98, which is an OmpA − E. coli K1. The expression of OmpA was equal in all these strains as analyzed by Western blotting of the outer membrane proteins of various mutants with an anti-OmpA antibody (Fig. 1B). In addition, the expression of OmpA on the surface of E. coli K1 also did not differ significantly as determined by flow cytometry (Fig. 1C). The growth curves of mutant E. coli K1 exhibited no differences, however, bacteria that lack OmpA expression grew slightly slower compared with OmpA + E. coli K1 (Fig.   1D). In addition, the levels of S-fimbriae and total polysaccharide content were also determined by dot blot analysis and thiobarbituric acid assay, respectively. As shown in Table 1, no significant differences were observed in the amounts of these two components produced among various strains. Similarly, the production of hemolysin was also unchanged in different mutants as determined by halo reaction on blood agar plates. It was previously shown that the lack of OmpA expression suppresses the level of type 1 fimbriation, which might contribute partially for the binding of E. coli K1 to endothelial cells (44).  Figure 1E). However, other mutants had retained the binding capacity similar to that of OmpA + E. coli K1. These results suggest that regions in loops 1, 2, and 4 of OmpA are involved in resisting serum bactericidal activity via increased binding to C4bp.

Different loop regions of OmpA are involved in the survival of E. coli K1 inside immune cells−Since
PMNs are the first line of defense against invading pathogens, we examined the effects of loop mutations in OmpA on the interaction of E. coli K1 with human PMNs. Interestingly, we observed that there were no significant differences in the entry of E. coli K1 mutants in PMNs ( Fig. 2A).
Bacteria with mutations in loops 1 and 3 of OmpA, however, exhibited slightly diminished multiplication up to 6 h post-infection. Similar results were also observed in mouse PMNs ( Fig.  S1A and B). Our previous studies have demonstrated that OmpA + E. coli K1 exploits dendritic cells (DCs) to manipulate their function, the most potent antigen presenting cells, for which OmpA expression is necessary (19). Therefore, we determined which regions of OmpA are required to enter and survive inside human DCs. No statistically significant differences were observed in the entry of OmpA mutant strains in DCs compared to wild type OmpA + E. coli K1 (Fig.   2C). However, mutations in loop 2 and loop 3 significantly prevented their survival in DCs (Fig.   2D). In contrast, mutations in loop 4 did not alter the survival, whereas mutations in loops 1 showed intermediate survival up to 6 h post-infection as seen in PMNs. Similar pattern of phagocytosis was also observed in mouse bone marrow derived DCs (BMDCs) (Fig. S1C and D). Furthermore, OmpA + E. coli K1 has been shown to persist and proliferate inside macrophages. Therefore we next examined the phagocytosis of OmpA mutant strains by mouse peritoneal macrophages. All the mutant strains of E. coli K1 were phagocytosed in similar numbers by macrophages (Fig. 2E). Of note, mutants 2a, 2b, 4a, 4b and 4c exhibited survival ability in mouse peritoneal macrophages similar to OmpA + E. coli K1. The number of intracellular bacteria recovered from macrophages increased by 6 h compared to 2 h post-infection same as that of PMNs and DCs, indicating proliferation of bacteria inside macrophages (Fig.  2F). In contrast, mutants 1a, 1b, 1c, 3a and OmpA − E. coli K1 were efficiently killed by macrophages.
Studies with mouse bone marrow derived macrophages showed similar results ( Fig. S1E and F). These data suggest that OmpA regions in loop 2 are critical to interact with PMNs, while loops 2 (amino acids) and 3 domains are necessary for interaction with DCs. In contrast, regions in loops 1 and 3 are important for survival inside macrophages.  (Fig. 3C). The brain bacterial titers were significantly higher in loop 4 mutant infected mice compared to OmpA+ E. coli K1 or loop 3 mutant infected animals ( Figure 3D). In contrast, brain cultures were sterile in mice infected with loop 1 and loop 2 mutants. In agreement with these findings, brain tissue damage markers lactate dehydrogenase (LDH) and malonaldehyde (MDA) were significantly increased whereas glutathione was decreased indicating oxidative stress in loop 4 mutants infected mice compared to animals infected with OmpA + E. coli K1 or loop 3 mutants ( Figure S2A-C). Since bacterial multiplication in tissues to reach a high level of bacteremia is important for crossing the blood brain barrier, we determined the bacterial load in other organs to examine whether the bacteria were trapped in these organs. High numbers of loop 4, loop 3, and OmpA + E. coli K1 were recovered from the intestine, liver, lung, spleen and kidneys of infected mice at 24 h post-infection ( Figure S3A-E). In contrast, very low numbers of bacteria were present in these organs in animals infected with loop 1 and loop 2 mutants. In agreement with these results, animals infected with loop 4 mutants showed greater disruption of the blood brain barrier compared to mice infected with loop 3 mutants and OmpA + E. coli K1 as demonstrated by Evan blue extravasation method ( Figure 3E). Interestingly, no disruption of the blood brain barrier was observed in loop 1 and loop 2 mutant infected animals. These results suggest that regions in loops 1 and 2 are responsible for surviving in tissues and/or blood to reach high-grade bacteremia, a prerequisite for the onset of meningitis.

Mutations in loop 4 cause exaggerated pathology in the brains of infected mice−Since loop 4 mutants infected mice succumbed earlier than wild
type E. coli K1 infected animals, we speculated that loss of epitopes in loop 4 might contribute to lethal immune response by surviving better in the animals. Therefore, we first examined the number and types of cells recruited to the brains of mice infected with various mutant strains of E. coli K1.
Interestingly, loop 1 and 2 mutant strains, which were cleared from the host faster than the wild type E. coli demonstrated more number of CD4+ and CD8+ T cells in brains, indicating that cellular immunity may be important for the clearance (Fig.  4A-F). In contrast, loop 4 mutants, showing higher virulence, recruited more microglia, B cells, macrophages, and granulocytes. Similarly, Loop 4 mutants upon infection, elicited higher production of TNF-α, IL-1β, IL-6 and IL-12 in blood compared to loop 3 mutants or OmpA + E. coli K1 ( Figure 5A-D). In contrast, loop 1 and loop 2 mutants showed significantly lower production of these cytokines. The entry of bacteria into the brain was accompanied with the activation of immune responses leading to generation of high levels of inflammatory cytokines, which constitutes the first line of self-defense against bacterial infections. However, exaggerated production of cytokines is a major cause of complications associated with meningitis and septic death. Therefore, we examined cytokine levels in brain tissues of mice infected with E. coli K1. We observed that levels of TNF-α, IL-1β, IL-6 and IL-12 were significantly high in the brains of mice infected with loop 4 mutants compared to animals infected with loop 3 mutants or OmpA + E. coli K1 (Figure S4A-D). Next, we determined the extent of brain damage in infected mice. Of note, loop 4 mutants caused more severe brain damage compared to OmpA + E. coli K1 infected pups.
Histopathology of brains from mice infected with loop 4 mutants displayed meningeal thickening and hemorrhage ( Figure 6). Severe leukocyte infiltration was observed in the cortex and white matter. Neurons exhibited pyknotic nuclei spanning entire regions of the dentate gyrus in the hippocampus of these mice. Severe hemorrhagic lesions were seen in the brain parenchyma of mice infected with loop 4 mutant strains of E. coli K1.
All these changes were present but less pronounced in OmpA + E. coli K1 or loop 3 mutant infected mice. However animals infected with 1a, 1b, 1c, 2a and 2b strains showed no brain damage similar to control uninfected animals. These results indicate that the pathophysiological manifestation of brain damage in animals infected with loop 4 mutants may be due to the entry of a greater number of bacteria, which in turn induced an exacerbated immune response.

DISCUSSION
In Gram-negative bacteria, the function of OmpA is thought to contribute to the structural integrity of the outer membrane (28)(29)(30). Recently, Power et al had identified two alleles of ompA, ompA1 and ompA2 (45). The nucleotide sequence differences of these alleles translate into several amino acid changes in the second and third extracellular loops. Our previous studies have shown that OmpA expression is very critical for the invasion of HBMEC by binding to Ecgp96 and that OmpA -E. coli K1 were 25-50 fold less invasive than the parental strain (13). Furthermore, we also demonstrated that two synthetic peptides corresponding to the sequences of loop 1 and loop 2 of OmpA significantly prevented the invasion of E. coli in HBMEC, albeit at high concentrations (15). Of note, the peptide sequence mimicking the region of loop 2 maps to the areas of OmpA that differ between ompA1 and ompA2 alleles (45). This indicates that specific subsets of OmpA exist which may be more or less invasive depending on the loop 2 amino acid sequence. The role of OmpA is not solely restricted to meningitic E. coli K1 binding to HBMEC. Enterohaemorrhagic E. coli (EHEC) of serotype O157:H7 utilizes OmpA in adhering to HeLa epithelial cells and Caco-2 colonic epithelial cells (46). OmpA also appears to be critical in adherence to plant surfaces, as an ompA mutant of E. coli O157 did not colonize alfalfa bean sprouts (47). In addition, OmpA, together with the Hek protein, has been shown to be critical for the invasion of colonic epithelial cells by meningitic E. coli (48). We have further demonstrated that OmpA expression is necessary for the interaction of E. coli K1 with neutrophils, dendritic cells, macrophages and complement protein C4bp (16)(17)(18)(19)(20). Nonetheless, no studies have been attempted to map out the regions of OmpA extracellular loops that interact with various host elements. Therefore, we sought to define the regions of E. coli K1 OmpA that interact with different immune cells and examine whether specific regions are necessary for the onset of neonatal meningitis.
Computer modeling studies on OmpA structure have shown that the extracellular loops are highly mobile and specific inter-loop interaction are critical for maintaining the three dimensional structure of OmpA (49) Therefore, deletion of any loop would distort the three dimensional structure of OmpA on E. coli K1. To avoid gross changes in the structure of OmpA, we generated mutants of OmpA in which three or four amino acids at a time have been exchanged for alanines. Although it was our aim to utilize "alanine scanning method" to sequentially change the amino acids to alanines in OmpA, some of the mutations were found to be lethal to the bacteria. Therefore, only nine OmpA mutants were generated and demonstrated to have no adverse effects to the phenotypic characteristics such as type 1 fimbriae and S-fimbriae expression, hemolysin and polysaccharide production. The structures were also modeled and their interaction with the receptor examined to ensure that their structures did not distort significantly (37). In addition, the invasion studies in HBMEC with these mutants demonstrated that regions of loops 1, 2, and 4 are important for invasion (37). Prediction of OmpA interaction with GlcNAc1-4GlcNAc epitopes of Ecgp96 based on the molecular modeling correlated >99% with experimental data validating our computer modeling studies (37). Since our previous studies demonstrated that greater amounts of C4bp in serum prevented the entry of E. coli K1 into HBMEC (18), we speculated that C4bp might be binding to the regions similar to those of OmpA receptor, Ecgp96. In this study, we demonstrated that loop regions 1, 2, and loop 4 are critical for binding to C4bp, a classical complement pathway regulator, which has been shown to be responsible for bacterial evasion of serum complement activity (16,17). This result also confirms our prediction that C4bp and Ecgp96 competes each other for the same binding site on OmpA. Of note, loop regions 1, 2, and 3 are necessary for survival in neutrophils similar to their role in interactions with HBMEC. Of note, we also observed that gp96, a homologue of Ecgp96 acts as a receptor on neutrophils for bacterial entry and survival inside the cells (unpublished data). In contrast, loops 2 and 3 played a major role in the survival of E. coli K1 in DCs. Although we have shown that E. coli infection induces CD47 expression in DCs, it is not clear at this time whether OmpA binds to CD47 or gp96 for survival in DCs (23). Furthermore, the survival of loop 1 and 3 mutants was significantly reduced in macrophages. We have demonstrated that OmpA interacts with Fc-γ receptor I (FcγRI) to bind to and enter the cells, which induced survival mechanisms in macrophages (21, 22 and unpublished data). Entrance through other receptors such as CR3 caused the bacterial degradation inside macrophages. Of note, greater expression of gp96 was also observed in macrophages, along with FcγRI and TLR2 upon infection with E. coli K1 both in vitro and in vivo (21,22). Thus, the binding affinity of OmpA to FcγRI could be greater than the binding to gp96 in macrophages. Interestingly in this study we observed that mutations in loop 4 induced elevated levels of immune response and influx of various immune cells into brain. Given a log increase in the number of bacteria in blood at 6 h post-infection, it is possible that loop 4 mutants may be avoiding initial serum killing and thus multiply faster. We have previously shown that neonatal serum kills E. coli K1 at or below 10 3 CFU/ml concentrations within 30 min in vitro (16). Since, loop 4 mutant showed no significant increase in multiplication in immune cells compared to wild type E. coli K1, it is likely that some serum complement components could not bind loop 4 mutants to kill the bacteria and thus avoid serum bactericidal activity initially.
In summary, our studies demonstrated the involvement of various loops of OmpA in the interaction with different cell types and proteins during the establishment of disease. Since loop 2 region appears to be participating in majority of these interactions and its sequence similarity with allele ompA2, which has been shown to present in pathogenic bacteria, targeting loop 2 regions for immunization would be a possible therapeutic option for prevention of this deadly disease.    which loop 4 mutations were incorporated showed higher brain bacterial titers compared to other strains (D). The blood-brain barrier permeability was assessed using the Evans blue extravasation technique as described in Experimental Procedures (E). The results were obtained from five independent experiments and the number of animals indicated is the total number of pups. Data represent means ± SD. *p <0.05 and + p <0.01 versus OmpA + E. coli K1 by two-tailed Student's t test.   Brains were harvested from E. coli K1 infected mice and subjected to H and E staining. Neutrophil infiltration (yellow arrow) in cortex and meninges in brains of mice infected with loop 3 mutants or OmpA + E. coli K1 was observed. These animals also showed infiltration of inflammatory cells evident in white matter (black arrow) and acute hemorrhage (yellow arrow) in the brain parenchyma. Hippocampus showed apoptosis of neurons (yellow arrow and inset) and inflammation (white arrow and inset). All these brain pathological changes were more severe and pronounced in mice infected with loop 4 mutants. Mutations in loop 1 and loop2 of OmpA rendered the E. coli K1 unable to cause brain damage in mice and therefore appeared similar to control uninfected mice. The results are representative of five independent experiments with 12 animals per group. Scale bars: 20 μM.