Adherence to, Invasion by, and Cytokine Production in Response to Serotype VIII Group B Streptococci

ABSTRACT The adherence to and invasion of the human epithelial cell line A549 by group B streptococcus (GBS) serotype VIII strains were compared with those of serotype III strains by a conventional method and the dynamic in vitro attachment and invasion system. Twenty GBS strains, including nine vaginal isolates and one invasive isolate each of serotypes III and VIII, were used in the conventional attachment and invasion assay. Adherence to and invasion of A549 cells by serotype VIII GBS strains were significantly greater (P < 0.0001) than those by serotype III strains for both the invasive strain and vaginal isolates. Cytokine production by A549 cells following stimulation with GBS serotypes III and VIII or their purified capsular polysaccharides (CPS) was measured. Serotype III strains stimulated significantly greater tumor necrosis factor alpha (TNF-α) (P < 0.0001) and interleukin-10 (IL-10) (P < 0.05) production than did serotype VIII strains. IL-8 production in response to serotype VIII was significantly higher (P < 0.001) than that in response to serotype III. TNF-α, IL-8, and IL-10 production was greater in A549 cells infected with GBS than in the untreated control cells. TNF-α production was significantly greater (P < 0.005) after stimulation with purified GBS serotype III CPS than after stimulation with serotype VIII CPS, a result similar to that after stimulation with whole GBS. IL-12 production by A549 cells was observed only in response to infection with GBS serotype III, resulting in the possibility of a greater TH1 response in serotype III GBS. These results suggest differences in immune responses to infection with GBS serotypes III and VIII.

Group B streptococci (GBS), or Streptococcus agalactiae, are a component of the normal flora colonizing the respiratory, gastrointestinal, and urologic tracts of humans. They are important pathogens that cause neonatal pneumonia, sepsis, and meningitis associated with high morbidity and mortality (8,42).
GBS are classified according to the presence of type-specific capsular polysaccharides (CPS), of which nine have been identified. GBS disease in the United States is caused primarily by types Ia, Ib, II, III, and V (12). Strains of serotypes IV, VI, VII, and VIII have rarely been isolated from patients in the United States (24,36). However, in many regions of the world, the serotypes that cause GBS infections are not restricted to those most prevalent in the United States (48). Several studies revealed the prevalence of GBS serotypes VI and VIII among pregnant women in Japan (13,21,30,48). Of particular interest is the fact that GBS serotype VIII, the serotype most frequently isolated from colonized women in Japan, is rarely associated with neonatal disease. Ongoing monitoring of the distribution of GBS serotypes is important for charting changes in serotype prevalence and for determining the components of an effective GBS vaccine. The differences in serotype distribution among various populations also may reflect differences in pathogenesis among the serotypes. We hypothesized that such differences between one of the dominant U.S. serotypes (serotype III) and one of the dominant Japanese serotypes (serotype VIII) might be present.
Adherence of bacteria and their invasion into host tissue is mediated through a complex series of events that involve changes in surface constituents (40). We examined adherence to and invasion of GBS serotype III and VIII strains into cultured A549 epithelial cells, using both a conventional method (41) and the dynamic in vitro attachment and invasion system (DIVAS) (16,25).
Innate proinflammatory responses to GBS infection that may contribute to respiratory pathology include the synthesis and release of cytokines, prostaglandins, and nitric oxide (20,22,26). Cytokines are also important mediators in host defenses against bacterial and viral infections (18). Cytokines are secreted proteins that function in intracellular signaling during inflammatory and immune processes (18). In general, proinflammatory cytokines, such as tumor necrosis factor alpha (TNF-␣), interleukin-1␣ (IL-1␣), IL-1␤, IL-6, IL-8, and IL-12, function as part of a cascade that orchestrates the recruitment and activation of inflammatory cells and the induction of effectors (18,44). TNF-␣ and IL-8 are produced by a variety of host cells upon exposure of the cells to a range of unrelated microbial components (7). IL-10 is produced by a variety of cells, including CD4 ϩ T cells, B cells, and macrophages. IL-10 possesses several characteristics and is a potent down-modulator of immune response (10,35,49,50). IL-12 is an integral immune regulator which promotes TH1 response while suppressing TH2 response (34). The low frequency of infection with serotype VIII might be associated with the ability of these strains to provoke the production of these cytokines. To address this possibility, we examined the cytokine response of human lung epithelial cells following stimulation with GBS serotypes III and VIII.
Cell line. The A549 cell line (ATCC CCL-185) was derived from a human lung carcinoma; it has many of the characteristics of type II alveolar epithelial cells, including surfactant production in culture, and has been used in our laboratory for invasion experiments (16). A549 cells were maintained in RPMI 1640 culture medium (GIBCO, Invitrogen, Grand Island, N.Y.) with 1% L-glutamine and 10% fetal bovine serum (GIBCO, Invitrogen) without antibiotics. The medium was replaced 1 day before the experiment. A549 cells suspended in 0.5 ml of RPMI 1640 with 10% fetal bovine serum were seeded into each well of 24-well culture plates and incubated for 4 days at 37°C with 5% CO 2 . By the beginning of the assay, the cell number had increased to approximately 1.5 ϫ 10 6 cells per well. Cell viability was determined by 0.4% trypan blue exclusion and hemocytometer counts.
Invasion and adherence assays. The attachment and invasion assays were done essentially as described by Rubens et al. (41), and DIVAS was performed as described previously (16), with the modifications indicated below. For the conventional assays, GBS were grown to an optical density at 650 nm (OD 650 ) of 0.3 to 0.5 in Todd-Hewitt broth (THB) (Becton Dickinson and Co., Sparks, Md.) and inoculated onto A549 monolayers for 2 h in each well of 24-well plates at 1.5 ϫ 10 6 CFU/well to achieve a multiplicity of infection of 1:1. The infected monolayers were then washed three times with phosphate-buffered saline (PBS) (ICN Biochemicals, Aurora, Ohio); 1 ml of fresh RPMI 1640 medium with 10% fetal bovine serum, 5 g of penicillin G (Sigma-Aldrich, St. Louis, Mo.) per ml, and 100 g of gentamicin (Sigma-Aldrich) per ml was added to each well, and plates were incubated in a CO 2 incubator for 2 h. Monolayers were rinsed five times with PBS, and A549 cells were detached by addition of 0.2 ml of trypsin-EDTA (GIBCO, Invitrogen), lysed by addition of 0.8 ml of 0.025% Triton X-100 (LabChem, Pittsburgh, Pa.), transferred quantitatively to microtubes, and mixed on a Vortex mixer for 1 min. Appropriately diluted aliquots in 0.025% Triton X-100 were plated on tryptic soy-5% sheep blood agar plates. The plates were incubated at 37°C overnight, and the number of GBS CFU recovered per well was determined. Each test was done in quadruplicate.
Bacterial adherence was determined as follows. GBS grown in THB to mid-log phase (OD 650 of 0.3 to 0.5) were inoculated at 1.5 ϫ 10 6 CFU/well onto A549 monolayers for 2 h in each well of 24-well plates. The infected monolayers were then washed five times with PBS, detached from the wells with 0.2 ml of trypsin-EDTA, lysed with 0.8 ml of 0.025% Triton X-100, transferred quantitatively to microtubes, and vortexed vigorously for 1 min. Aliquots diluted in 0.025% Triton X-100 were plated on tryptic soy-5% sheep blood agar plates. The plates were incubated at 37°C overnight, and GBS CFU were counted. The number of attached GBS was calculated as total (attached and invaded) CFU minus invaded CFU. Each test was done in quadruplicate, and the number of CFU recovered per well was determined.
DIVAS. DIVAS was developed to study microbe-host cell interactions with bacteria held at steady-state growth in a chemostat under defined environmental and nutritional conditions (16,25). Bacteria held at the steady state of growth are perfused slowly over an established monolayer of tissue culture cells as described below. In brief, a 1-liter chemostat vessel (Applikon, Foster City, Calif.) with a working volume of 0.5 liter was used to grow GBS in chemically defined medium with glucose limitation at a mass doubling time (t d ) of 1.8 h (fast growth). The temperature (37°C), pH (pH 7.3), and mixing rate (400 rpm) were maintained at constant levels throughout the experiments. Levels of dissolved oxygen (DO 2 ) were monitored with a polarographic DO 2 probe and controlled by the addition of a mixture of compressed nitrogen and oxygen. All growth parameters (pH, temperature, mixing rate, media inflow rate, and DO 2 ) were computer controlled with a BioController (model ADI 1030; Applikon).
Attachment and invasion experiments were performed with A549 respiratory epithelial cells grown in 60-mm-diameter culture dishes. Confluent cells were washed twice with 3.0 ml of PBS prior to use in DIVAS. GBS from the chemostat were perfused over the A549 cells by using a plate holder apparatus with inlet and outlet ports to allow for perfusion of bacteria (25). The apparatus was placed on a rotary shaker at 20 rpm in a 37°C incubator, and perfusion of GBS proceeded for 2 h, after which the plates were disconnected from the outflow line of the chemostat. Cell monolayers were washed with PBS, extracellular GBS were killed with antibiotics, eukaryotic cells were disrupted, and internalized GBS were quantified by standard plate counts as described previously (16,25). SMU007 (serotype III) and SMU029 (serotype VIII) GBS were used for DIVAS assay. Invasion and adherence with DIVAS were performed with GBS held at a t d of 1.8 h and 12% DO 2 , conditions that result in high levels of invasion (16). Cytokine assays were also performed twice on filtered outflow supernatant fluid samples obtained at 0, 1, and 2 h after GBS passed over the A549 cell monolayer.
Sampling method for cytokine assays. GBS were grown to an OD 650 of 0.3 to 0.5 in THB. GBS were inoculated at 1.5 ϫ 10 6 CFU/well onto A549 monolayers in each well of 24-well plates at an MOI of 1:1. The supernatant fluid was removed after 2 h of contact with the bacteria. The supernatant fluids were clarified by centrifugation and were filtered (0.22-m-pore-size filter) for clarification.
Effect of stimulation with GBS CPS on TNF-␣ production. TNF-␣ production by A549 cells stimulated with serotype III and VIII GBS CPS at 2, 5, and 10 g/ml was investigated. GBS serotype III and serotype VIII CPS were purified as previously described (47,48). Each test was done in triplicate, and cytokine assays were conducted twice in duplicate. Because of the potential for the presence of trace amounts of endotoxin in the purified CPS, each CPS was exposed to polymyxin B (2, 5, and 10 g/ml) (Sigma-Aldrich) at stimulation. Each test was done in quadruplicate, and cytokine assays were conducted twice in duplicate.
Statistical analyses. The significance of GBS invasion for two different variables was determined by the unpaired t test with the Welch correction. These analyses were performed with Instat version 3.0a (GraphPad, San Diego, Calif.). P values of Ͻ0.05 were considered significant.

Adherence and invasion.
Adherence to and invasion into A549 cells by serotype VIII GBS was significantly higher than those by serotype III GBS (Table 1) both for the invasive isolate (P Ͻ 0.0001) and for vaginal isolates (P Ͻ 0.0001). For serotype VIII, a significant difference (P Ͻ 0.005) between the invasion into A549 cells by invasive isolate NKMC1998 and vaginal isolates was seen, whereas for serotype III, no significant difference (P Ͼ 0.05) between invasive and vaginal isolates was noted. For both serotypes, there was no significant difference (P Ͼ 0.05) in the adherence of invasive and vaginal isolates to A549 cells.
To determine the invasiveness of GBS held at a steady state of growth and perfused over A549 cells, we performed DIVAS with two selected strains of GBS: serotype III SMU007 and serotype VIII SMU029. A mean (Ϯ standard deviation) of 17 Ϯ 2.8 GBS serotype III SMU007 cells adhered per A549 cell, compared with 24 Ϯ 0.7 GBS serotype VIII SMU029 cells (P Ͻ 0.05) (Fig. 1).
GBS serotype VIII SMU029 grown in continuous culture invaded A549 cells more readily than did GBS serotype III SMU007 (Fig. 1). The CFU of GBS serotypes III and VIII grown at a t d of 1.8 h with 12% DO 2 that invaded A549 cells were 672 Ϯ 56 and 7,900 Ϯ 1,800, respectively (P Ͻ 0.0001). Cytokine production. Secretion of TNF-␣, IL-8, and IL-10 by A549 cells in response to GBS infection was significantly increased compared with secretion by the untreated control group. There was a significant difference between TNF-␣ production by serotypes III and VIII for both invasive and vaginal isolates (P Ͻ 0.0001 and P Ͻ 0.0001 for invasive and vaginal isolates, respectively) (Fig. 2). There was a significant difference in IL-8 production between serotypes III and VIII for both invasive and vaginal isolates (P Ͻ 0.001 and P Ͻ 0.001 for invasive and vaginal isolates, respectively) (Fig. 2). IL-10 production by A549 cells was similar to that of TNF-␣. There was a significant difference in IL-10 production in response to serotypes III and VIII for both invasive (P Ͻ 0.05) and vaginal (P Ͻ 0.01) isolates (Fig. 2). IL-12 production by A549 cells in response to GBS infection was observed for serotype III, while it was not observed for serotype VIII (Fig. 2).
Since the flow of chemically defined medium in the DIVAS assay is continuous, cytokines do not accumulate but are diluted in this open-flow system over time; thus, the level of cytokine production was relatively low in all samples. Nevertheless, TNF-␣ production in response to serotypes III and VIII differed significantly (P Ͻ 0.0001 and P Ͻ 0.0001 at 1 and 2 h, respectively) (Fig. 3). Also, there was a significant difference in IL-8 production in response to serotypes III and VIII (P Ͻ 0.0005 and P Ͻ 0.0001 at 1 and 2 h, respectively) (Fig. 3). On the other hand, IL-10 production could be detected only for serotype III GBS at 2 h and was below the detection limit for serotype VIII (Fig. 3).
Effect of stimulation with GBS CPS on TNF-␣ production. TNF-␣ production by A549 cells increased in a dose-dependent manner with both serotypes after stimulation with CPS (Fig. 4). There was a significant difference in TNF-␣ production in response to CPS of serotypes III and VIII at all concentrations tested (P Ͻ 0.005, P Ͻ 0.0001, and P Ͻ 0.0001 at CPS concentrations of 2, 5, and 10 g/ml, respectively) (Fig. 4). TNF-␣ production by A549 cells was inhibited by only 26.2% (serotype III CPS) and 21.5% (serotype VIII CPS) following stimulation of A549 cells by CPS with polymyxin B. TNF-␣ production by A549 cells differed significantly (P Ͻ 0.0005) following stimulation by either serotype III CPS or serotype VIII CPS.

DISCUSSION
Since the early 1980s, GBS strains of serotypes VI and VIII have been prevalent in Japan among isolates colonizing preg- a Adherence to and invasion into A549 cells were significantly, higher for serotype VIII GBS strains than for serotype III strains, both for the invasive strain and for vaginal strains (P Ͻ 0.0001 and P Ͻ 0.0001 for CFU of invasive and vaginal strains, respectively). There was a significant difference between invasion into A549 cells by invasive strain NKMC1998 and by vaginal strains of serotype VIII (P Ͻ 0.005), but there was no significant difference between invasion by invasive and vaginal strains for serotype III (P Ͼ 0.05).
b Number of adherent CFU/number of CFU in inoculum. c Number of intracellular CFU/number of CFU in inoculum.
nant women, with serotype VIII accounting for 25 to 36% of all isolates (21,30). Despite this high carriage rate, the frequency of early-onset disease due to serotype VIII among Japanese neonates is low (14). In Japan, a nationwide surveillance of early-onset neonatal GBS infection conducted from 1983 to 1997 found that the most frequent serotypes among 224 strains examined were serotype III (38.4%), serotype Ia (17.0%), serotype Ib (8.5%), serotype VI (8.5%), serotype II (6.7%), and serotype VIII (4.9% [11 cases]) (14). Recent studies have shown that women colonized with GBS serotypes Ia, Ib, II, III, and V had significantly higher concentrations of IgG specific for the CPS of the colonizing serotype in serum compared with women not colonized with GBS (2,4,11). We hypothesized that a difference in the host response to serotype III and VIII strains may account for this disparity and that this difference may be associated with their ability to invade host cells. The results of our in vitro study support the hypothesis that serotype VIII strains are more invasive than strains of serotype III. Factors that may account for differences between these two serotypes are numerous and may involve fine structural features of the CPS or expression of surface proteins associated with attachment and invasion (31). Significant differences in attachment and invasion of sero-type VIII strains and serotype III strains were independent of the assay (conventional or DIVAS) used. Of the two vaginal isolates tested in DIVAS, higher numbers of GBS serotype VIII than of GBS serotype III attached to and invaded A549 cells, a surprising finding because the number of cells of sero- The means of TNF-␣ production following stimulation with serotypes III and VIII are 75.1 and 46.1 pg/ml, respectively. The means of IL-8 production for serotypes III and VIII are 541.2 and 677.4 pg/ml, respectively. The means of IL-10 production for serotypes III and VIII are 11.9 and 9.7 pg/ml, respectively. The mean of IL-12 production for serotype III is 1.74 pg/ml. Error bars indicate standard deviations. . The difference in invasiveness between an invasive and a colonizing GBS may be due to the difference in expression of factors specifically involved with the invasion process or to differences in metabolism and nutrient requirements. An understanding of mechanisms responsible for these differences awaits further experimentation. Host immune factors have an obvious and important role in disease progression. Matsubara et al. reported that women colonized with serotype VIII had significantly higher titers of serum antibodies to serotype VIII CPS than did women who were not colonized (29). It is possible that entry into epithelial cells might enhance the antibody response to GBS, possibly through better antigen processing, thus diminishing the rate of invasive infection caused by serotype VIII GBS compared with the rate of colonization by this serotype. We have hypothesized that in some circumstances, entry into epithelial cells may serve as a host defense mechanism, since GBS do not to replicate intracellularly (41) and may be expelled from the host as the epithelial cells are shed. Increased systemic concentrations of proinflammatory cytokines, including TNF-␣, have been correlated with septic shock and mortality (6,46). The ability of GBS to induce proinflammatory cytokines has been documented for human leukocytes (45,47) and in rodent infection models (23,28). In those models, TNF-␣ plays a central role in mediating mortality (28,45). We found that the production of TNF-␣ by A549 cells following stimulation with serotype III GBS was significantly higher than that following stimulation with serotype VIII, although adherence to and invasion of A549 cells were lower for serotype III than for serotype VIII. Therefore, TNF-␣ expression may be associated with the major virulence factor (i.e., CPS antigen) of GBS and may account for the greater virulence of serotype III strains. The fact that production of IL-12 was observed in response to serotype III and not for serotype VIII infection might lead to more antibody production for women colonized by serotype VIII. The elucidation of cellular mechanisms involved in GBS-induced cytokine production may also be important for the development of effective therapeutic strategies (27). These data are consistent with the hypothesis that greater invasion by GBS serotype VIII and lower TNF-␣ production tend to drive a TH2 response, resulting in higher levels of protective antibody.
IL-8 is a potent neutrophil chemotactic stimulus (33) that further acts to up-regulate the expression of adhesion molecules on the neutrophil surface (5), to promote their transendothelial migration (15), and to stimulate an oxidative burst and release of lysosomal enzymes (37). An elevated level of circulating IL-8 is a strong predictor of bacterial infection (9), and marked increases in levels of IL-8 have been detected in sputum, bronchoalveolar lavage fluid, and uterine endometrial lavage fluid from patients with various infectious diseases (32,39,43). IL-8 is also the major neutrophil chemoattractant produced by lung epithelial cells after experimental infection (19). In the present study, we found that GBS induced IL-8 release from A549 lung epithelial cells in a cell surface adhesion-, cell invasion-, and time-dependent fashion (data not shown). The greater amount of IL-8 may lead to recruitment of more inflammatory cells, which may lead to the more rapid elimination of GBS serotype VIII strains.
It is interesting that we detected IL-10 as well as TNF-␣ and IL-8 production by A549 cells cultured in the presence of GBS. IL-10 has been shown to inhibit the production of many proinflammatory cytokines, including IL-1, IL-6, TNF-␣, IL-8, granulocyte-monocyte colony-stimulating factor, and granulocyte colony-stimulating factor; to down-regulate activity by macrophages/monocytes and polymorphonuclear cells (1,3,38); and to inhibit leukocyte chemotaxis (17). The presence of IL-10 may represent a cellular response to the increasing levels of TNF-␣, since this has been described as an autoregulatory mechanism to decrease the production of TNF-␣.
Production of TNF-␣ was significantly greater after stimulation by serotype III CPS than after stimulation by serotype VIII CPS, a result similar to that for stimulation by whole bacteria. This indicates that production of TNF-␣ is also dependent on bacterial surface components such as CPS.
It is clear that serotype III isolates provoke a more robust inflammatory response while at the same time potentiating the down-regulation of the immune response via IL-10. The diminished IL-10 response and the absence of an IL-12 response to serotype VIII GBS may result in a more robust production of protective antibody via a TH2 route. Continued research into the differences in the pathogenesis of different strains of the nine GBS serotypes and between colonizing and invasive isolates will provide a more complete understanding of factors important for GBS pathogenesis. Future studies will address the possibility of a predominant TH2 response with concomitant antibody production in women colonized with serotype VIII GBS.