Alpha C Protein of Group B Streptococcus Binds Host Cell Surface Glycosaminoglycan and Enters Cells by an Actin-dependent Mechanism*

Group B Streptococcus (GBS) colonizes mucosal surfaces of the human gastrointestinal and gynecological tracts and causes disease in a wide range of patients. Invasive illness occurs after organisms traverse an epithelial boundary and enter deeper tissues. Previously we have reported that the alpha C protein (ACP) on the surface of GBS mediates GBS entry into ME180 cervical epithelial cells and GBS translocation across layers of these cells. We now demonstrate that ACP interacts with host cell glycosaminoglycan (GAG); the interaction of ACP with ME180 cells is inhibited if cells are pretreated with sodium chlorate, an inhibitor of sulfate incorpora-tion, or with heparitinases. The interaction is also inhibited in the presence of soluble heparin or heparan sulfate or host cell-derived GAG. In addition, ACP binds soluble heparin specifically in inhibition and dot blot assays. After interaction with host GAG, soluble ACP enters ME180 cells and fractionates to the eukaryotic cell cytosol. These events are inhibited in cells pretreated with cytochalasin D or with Clostridium difficile toxin B. These data indicate that full-length ACP interacts of of the BSA-treated control cell population.Inhibition assays performed but prior to incubation with cells treated with reagents, m incubated ature h), g/ml, for cytochalasin g/ml, sucrose brefeldin A Clostridium toxin or ng/ml, incubated at 37 °C for 24 h). The incubation conditions were selected based upon literature reports and upon pilot studies in which cell morphology was observed by light microscopy periodically during incubation to ensure that pretreatment did not produce any loss of monolayer integrity. Additional studies were performed by preincubating Alex-aFluor-labeled ACP or BSA with varying concentrations of inhibitor chondroitin A, B, and C, hyaluronic N-acetylglucosamine, GBS type Ia or cell-derived GAG fo r 1 h at 4°C prior to this to ME180 cells and incubating as


Group B Streptococcus (GBS) colonizes mucosal surfaces of the human gastrointestinal and gynecological tracts and causes disease in a wide range of patients. Invasive illness occurs after organisms traverse an epithelial boundary and enter deeper tissues. Previously we have reported that the alpha C protein (ACP) on the surface of GBS mediates GBS entry into ME180 cervical epithelial cells and GBS translocation across layers of these cells. We now demonstrate that ACP interacts with host cell glycosaminoglycan (GAG); the interaction of ACP with ME180 cells is inhibited if cells are pretreated with sodium chlorate, an inhibitor of sulfate incorporation, or with heparitinases. The interaction is also inhibited in the presence of soluble heparin or heparan sulfate or host cell-derived GAG. In addition, ACP binds soluble heparin specifically in inhibition and dot blot
assays. After interaction with host GAG, soluble ACP enters ME180 cells and fractionates to the eukaryotic cell cytosol. These events are inhibited in cells pretreated with cytochalasin D or with Clostridium difficile toxin B. These data indicate that full-length ACP interacts with ME180 cell GAG and enters the eukaryotic cell cytosol by a mechanism that involves Rho GTPase-dependent actin rearrangements. We suggest that these molecular interactions drive ACP-mediated translocation of GBS across epithelial barriers, thereby facilitating invasive GBS infection.
Streptococcus agalactiae (Group B Streptococcus, GBS) 1 has long been recognized as an important cause of infection in pregnant/peripartum women and neonates. A frequent colonizer of the human gastrointestinal and gynecological tracts, GBS has been noted more recently to cause a range of invasive syndromes in non-pregnant adults. Most commonly these pa-tients have comorbid conditions, including malignancy, diabetes, and renal disease (1), that may predispose to bacterial invasion because of a loss of epithelial barrier protection in a chronically colonized site such as the rectum, vagina, cervix, urethra, skin, or pharynx. The molecular basis of the interaction between GBS and epithelial cells remains poorly understood.
We have reported that the alpha C protein (ACP) on the surface of GBS interacts with epithelial cells. Expressed by many serotype Ia, Ib, and II GBS strains, ACP is the prototype for a family of Gram-positive surface proteins, the alpha-like proteins (Alps). Found on most GBS strains and some Enterococcus and group A Streptococcus strains, Alps share considerable sequence homology and common structural elements, including an N-terminal region, a series of tandem repeats of ϳ80 amino acids each, and a C-terminal region containing a cellwall anchor LPXTG motif common to several Gram-positive species. Despite the fact that these proteins may vary in size due to gene truncation within the repeat region (2), Alps elicit protective antibody in both adult and neonatal mouse models of GBS sepsis (3). In a neonatal mouse model of disease, deletion of the gene encoding ACP attenuates the virulence of GBS, with significant survival differences during the first 24 h of infection (4). In vitro, the ACP-deficient mutant both enters ME180 human cervical epithelial cells and transcytoses across layers of these cells less effectively than the wild-type parent strain (5). Recombinantly derived soluble protein representing the N-terminal region of ACP inhibits the entry of wild-type GBS into these cells and inhibits the association of full-length soluble ACP with these cells.
We sought to identify the host cell surface receptor interacting with soluble Alps, to determine whether soluble Alps enter eukaryotic cells, and, if so, to delineate the mode of entry and the intracellular destination of the protein after internalization. Because host cell surface glycosaminoglycans (GAGs) mediate internalization of other organisms (6 -11), we hypothesized these molecules might mediate Alp interactions with ME180 cells. Based on the data described above and the Nterminal region sequence similarities among Alps (12), we hypothesized that the binding of Alp N-terminal domains to a host cell surface GAG might trigger uptake of the ligand and attached structures into cell vacuoles.

Strains/Plasmids/Cells
GBS strain A909 is the prototype serotype Ia/C (␣ ϩ , ␤ ϩ ) strain (13). A909 genomic DNA contains a copy of the bca gene that encodes a nine-repeat ACP (14). The strain was grown in Todd Hewitt Broth (Difco). Escherichia coli strain BLR (Novagen) was used to express 1-repeat and 9-repeat alpha C proteins, as well as the 9-repeat region (without N-or C-terminal regions), and E. coli strain BL21(DE3) (Novagen) was used to express the N-terminal region of ACP from pDEK14 (15); constructs were designed and proteins were expressed and purified as described in Refs. 5 and 16. Soluble Alp3 was produced using PCR primers 5Ј-CGGGATCCTCTACAATTCCAGGGAGT-3Ј and 5Ј-GGAATTCTTAATCCTCTTTTTTCTT-3Ј to amplify the Alp3 gene (12), minus its signal sequence, from GBS serotype VIII strain H4A0158. The PCR product was cloned into pCR2.1 (Invitrogen), and the insert was excised with BamH1 and EcoR1 and cloned into the pTrcHisA expression vector (Invitrogen). The resulting construct was transformed into E. coli strain BL21(DE3). The protein was expressed and purified as described previously (5). Beta C protein was prepared as described in Ref. 17.

Flow Cytometry
Fluorescent Labeling of Proteins-An AlexaFluor 488 protein labeling kit (A-10235; Molecular Probes) was used to conjugate AlexaFluor 488 dye to bovine serum albumin (BSA) and Alp products, according to the manufacturer's instructions as described previously (5).
Cell Staining-ME180 cells were grown to monolayer confluence in 6-well plates with 2 ml of RPMI 1640 (Invitrogen), including 10% FCS (Invitrogen) and 1% penicillin/streptomycin (Invitrogen). The day prior to the assay, the medium was replaced with 1 ml of fresh medium, and the cells were incubated overnight at 37°C with 5% CO 2 . The next day AlexaFluor-labeled Alp product or BSA or mouse monoclonal antibody to known ME180 surface components CD46 (BD Pharmingen International) or epidermal growth factor receptor (EGFR, Sigma) was added to the wells for a final concentration of 0.1 M (18,19). The 6-well plates were then incubated at 37°C with 5% CO 2 for up to 24 h. The medium was removed from the wells, and the monolayers were washed three times with 1 ml of PBS to remove nonadherent proteins. Fluorescein isothiocyanate-labeled goat anti-mouse secondary antibody (Sigma) was added to wells treated with CD46-specific and EGFR-specific antibodies, and cells were incubated for 1 h then washed with PBS. 350 l of trypsin-EDTA (0.25% trypsin, 1 mM EDTA-4 Na, Invitrogen) was added to the wells, and plates were incubated for 10 min at 37°C. Cells were detached by repeat pipetting and harvested by centrifugation at 650 rpm (50 x g) for 8 min. Cells were washed with 1 ml of PBS and resuspended in 0.1 ml of 2% paraformaldehyde in PBS, then incubated at 4°C overnight. The samples were washed with 1 ml of PBS to remove the fixative, resuspended in 0.4 ml of PBS, filtered through a cellstrainer cap (Falcon), and analyzed by flow cytometry on the MoFLo (Cytomation) machine. The cell population of interest was identified by using the AlexaFluor-labeled BSA sample to define nonspecific staining and/or autofluorescence levels. Positive staining was defined as a fluorescent signal greater than that of 98.5% of the BSA-treated control cell population.
Inhibition assays were performed as described above, but prior to incubation with soluble proteins, cells were treated with reagents, including sodium periodate (Sigma, 10 mM, incubated at room temperature for 1 h), tunicamycin (Sigma, 10 g/ml, incubated at 37°C for 72 h), heparitinases (Seikagaku, 10 milliunits/ml, incubated at 37°C for 3 h), cytochalasin D (Sigma, 2.5 g/ml, incubated at 37°C for 2 h), chlorpromazine (Sigma, 25 g/ml, preincubated at 37°C for 30 min), hypertonic sucrose (0.3 M, preincubated at 37°C for 1 h), brefeldin A (Sigma, 5 M, preincubated at 37°C for 30 min), or Clostridium difficile toxin B (List Biological Laboratories, Inc., Campbell, CA, 10 or 100 ng/ml, incubated at 37°C for 24 h). The incubation conditions were selected based upon literature reports and upon pilot studies in which cell morphology was observed by light microscopy periodically during incubation to ensure that pretreatment did not produce any loss of monolayer integrity.
Additional inhibition studies were performed by preincubating Alex-aFluor-labeled ACP or BSA with varying concentrations of inhibitor (heparin sodium, chondroitin sulfates A, B, and C, heparan sulfate (all purchased from Sigma), hyaluronic acid, N-acetylglucosamine, GBS type Ia polysaccharide (kindly provided by Dr. Dennis Kasper and Barbara Reinap), or cell-derived GAG (see below)) for 1 h at 4°C prior to adding this mixture to ME180 cells and incubating as described above.

Confocal Microscopy
ME180 cervical epithelial cells were grown to confluence on glass coverslips in 6-well plates in 2 ml of RPMI 1640 medium with Lglutamine containing 10% FCS (Invitrogen) and 1% penicillin/streptomycin (Invitrogen). The night before the experiment, the medium was removed and replaced with 1 ml of fresh media. The following morning, AlexaFluor 488-labeled protein (ACP or BSA) at 0.5 M was added. The proteins were incubated with ME180 cells at 37°C for varying periods up to 4 h. The monolayers were washed three times with 1 ml of PBS, then treated with 0.25 ml of 2% paraformaldehyde fixative in PBS, incubated overnight at 4°C, and washed again with 1 ml of PBS. To mark the cell surface, the monolayers were then incubated with mouse monoclonal EGFR-specific antibody for 1 h, washed, and treated with a rhodamine-labeled anti-mouse secondary antibody (Sigma) for 1 h, then washed again. The coverslips were then mounted on glass slides with Fluorotec mounting media (Accurate Chem., Westbury, NY). The slides were reviewed by confocal microscopy (Bio-Rad MRC-1024/MP multiphoton microscope, using a multiline krypton/argon laser for standard confocal microscopy, interfaced with a Zeiss Axiovert microscope). All images were viewed at the same gain value. Inhibition assays were performed similarly, except cells were preincubated (as described above) with reagents that included cytochalasin D and brefeldin A.

Inhibition Assay
Modeled after enzyme-linked immunosorbent assay methods (17), heparin-albumin-biotin (Sigma) was added at 200 g/ml along with serially diluted concentrations of unlabeled heparin from 0 -50 mg/ml in incubation buffer (10 mM PBS with 0.05% Brij (Sigma Diagnostics)/ 2.5% newborn calf serum (Whittaker Bioproducts)), to wells of a microtiter plate (Nunc-immuno plate; American Bioanalytical) coated with 100 l of soluble ACP at 20 g/ml in 0.1 M carbonate buffer (pH 9.8). Absorbance at 405 nm (A 405 ) was measured after incubation with alkaline phosphatase-conjugated avidin (Pierce Endogen) diluted 1:3000 in incubation buffer and phosphatase substrate (Sigma). Optimal concentrations of ACP and biotinylated heparin were determined initially by checkerboard.

GAG Preparation from ME180 Cells (Modified from Ref. 20)
ME180 cells were grown to confluence in 100 mm ϫ 20 mm polystyrene culture dishes (Corning), then incubated at 37°C overnight in 4 ml of 6ϫ Pronase solution (1 mg/ml Pronase (Calbiochem), 0.24 M sodium acetate, 1.92 M sodium chloride, pH 6.5) and 20 ml of culture media. The following morning, the resulting material was spun at 3000 rpm for 10 min, and supernatant was filtered, diluted 1:1.5 with water to diminish the salt concentration, and adjusted to pH 6.5 with 0.01% Triton X-100. The diluted supernatant was loaded onto a 1-ml DEAE-Sephacel column (Amersham Biosciences) that had been pre-equilibrated with 12 ml of 0.25 M sodium chloride, 20 mM sodium acetate, 0.01% Triton X-100, pH 6.0. The column was washed with 30 ml of 0.25 M sodium chloride, 20 mM sodium acetate, 0.01% Triton X-100, pH 6.0, then GAG was eluted with 5 ml of 1 M sodium chloride in 20 mM sodium acetate, pH 6.0. Ethanol (10 ml per 2.5-ml eluate) was added to the eluate, and the GAG was allowed to precipitate at 4°C overnight. The following morning, precipitate was spun for 10 min at 3000 rpm, and the pellet was washed with 1 ml of 70% ethanol, centrifuged for 7 min at 3000 rpm, and dried. The resulting material was resuspended in PBS for further studies.
GAG concentration was measured by our kind collaborators Dr. Robert Rosenberg and Andre Love using Beckman's chromatographic equipment. Purified GAG was acid hydrolyzed with 6 N HCl. Automated o-phthaldialdehyde (OPA, Sigma) derivatization was performed with the 508 autosampler that was programmed to draw 31 l of borate buffer, 4 l of OPA, plus 5 l of GAG sample and mix into the destination vial prior to autoinjection. Chromatographic separations were performed through a MAX-RP C12 column (Phenomenex). The mobile phase flow rate was 0.8 ml/min, and a binary gradient was employed where mobile phase A was 0.1 M sodium phosphate monobasic, pH 6.4, mobile phase B was mobile phase A, methanol, and tetrahydrofuran (30:70:3, v/v). For detection of the OPA-labeled sugars, the excitation and emission wavelengths were set at 340 and 450 nm, respectively. Titrations of glucosamine and Nor-leucine were used to generate an external standard curve from which GAG quantitations were obtained. A known amount of Nor-leucine, as the internal standard, was added to samples prior to hydrolysis. The chromatographic areas of GAG were calculated from the external standard curve and corrected with the internal standard. The amount of glucosamine in the hydrolyzed GAG sample was then used to calculate the concentration of GAG in the original sample as follows: grams of glucosamine ϫ 450 (avg. molecular weight of disaccharide)/179 (average molecular weight of glucosamine) ϭ grams of GAG.

Cellular Fractionation Techniques
Digitonin Treatment (Modified from Ref. 21)-Confluent ME180 cells were incubated with 0.5 M soluble ACP at 37°C for 1.5 h. Cells were washed to remove unbound protein and then treated with 1% digitonin (Sigma), which binds membrane sterols, resulting in selective solubilization of the eukaryotic plasma membrane. The cells were detached using a rubber policeman and centrifuged at 20,000 ϫ g to separate cytosolic components (supernatant) from cellular material (pellet). The samples were treated with methanol and chloroform to extract proteins, then suspended in SDS sample buffer, subjected to SDS-PAGE, and evaluated by Western blot using 4G8, a monoclonal antibody to the ACP repeat region (3).
Mechanical Fractionation (Modified from Ref. 22)-Confluent ME180 cells were incubated with ACP as described above, then detached from flasks by trypsin treatment, washed, and passed through a 27-gauge needle to lyse the cells. The resulting material was spun sequentially at 2500 rpm to pellet intact cells, at 8000 rpm to pellet nuclei, mitochondria, and other large organelles, and finally at 40,000 rpm to pellet lysosomal/endosomal compartments. The pellets were solubilized in 1% SDS. The final supernatant and solubilized pellets were treated with methanol and chloroform to extract proteins as described above, then evaluated by Western blot with monoclonal antibody recognizing ACP.

Dot Blots
Modeled after an immunoblot assay, protein binding to heparin was studied with a modified dot-blot technique (23). Proteins (1 g of each) were applied to a nitrocellulose membrane, which was then blocked with 5% skim milk for 1 h. The membrane was then incubated with heparin-albumin-biotin (Sigma) at 0.05 mg/ml for 1 h, followed by alkaline phosphatase-conjugated avidin (Pierce Endogen) for 1 h before washing and developing.

Bacterial Internalization Assays
These were performed as described in a previous study (5). For inhibition studies, these assays were modified as follows: ME180 cells were preincubated with individual ACP or Alp3 inhibitors for 1 h at 37°C with 5% CO 2 prior to adding GBS.

RESULTS
Eukaryotic Cell GAG Interacts with ACP-The ability of ACP to associate with cells despite the trypsin treatment used for detachment of cells prior to flow cytometry analysis (5), in combination with the importance of host cell surface GAG molecules in mediating entry of many pathogens (6 -11), led us to hypothesize that ACP might bind to cell surface GAG. In support of this hypothesis, cell staining with labeled ACP was diminished in ME180 cells pretreated with sodium periodate (which oxidizes surface carbohydrates) or with tunicamycin (which inhibits protein glycosylation) (data not shown). The following studies demonstrate more specifically that cellular GAG interacts with ACP.
Sodium Chlorate Treatment of ME180 Cells Decreases Cellular ACP Association-To prevent sulfation of glycoproteins and carbohydrates (24), we treated confluent ME180 cells with sodium chlorate then added labeled ACP. This treatment diminished the cellular accumulation of ACP by 86.4 -94% as measured by flow cytometry analysis (Fig. 1). These data indicate that a sulfated host cell structure, such as GAG, interacts with ACP.
Heparitinase Treatment of ME180 Cells Decreases Cellular ACP Association-Treatment of confluent ME180 cells with heparitinases prior to incubation with ACP diminished the cellular accumulation of ACP by 98% compared with untreated cells in flow cytometry analysis (Fig. 1). In contrast, pretreatment of ME180 cells with chondroitinase ABC (Sigma) did not diminish staining. These data demonstrate that ACP recognizes a cellular GAG that is altered by heparitinase treatment.
Soluble GAG Inhibits Association of ACP with ME180 Cells-To investigate the type of cell surface GAG involved, we preincubated labeled soluble ACP with varying concentrations of soluble heparin (Sigma), then added this mixture to confluent ME180 cells. Flow cytometry analysis of the cells shows ACP accumulation diminished in proportion to the concentration of soluble heparin (Fig. 2). Specifically, 500 g/ml heparin reduced ACP staining by 59.2%. At similar concentrations, heparan sulfate diminished ACP staining by 58.2%, whereas a mixture of chondroitin sulfate A and chondroitin sulfate C (Sigma C6737, ϳ57% type A and 43% type C) diminished ACP staining by 33.3%. Similar concentrations of other sulfated and non-sulfated carbohydrates, including chondroitin sulfate B, chondroitin sulfate C, hyaluronic acid, GBS type Ia capsular polysaccharide, and N-acetylglucosamine, showed no substantive effect on the interaction of ACP with ME180 cells. These data indicate specificity in the interaction between ACP and heparin/heparan sulfate. Soluble ACP Specifically Binds Soluble Heparin-To further evaluate the degree of specificity with which heparin binds to ACP, we performed an inhibition assay, coating a microtiter plate with ACP (20 g/ml), then incubating with 200 g/ml heparin-albumin-biotin (Sigma) and unlabeled heparin in concentrations ranging from 0 to 25 mg/ml. After washing, bound biotinylated heparin was detected using alkaline phosphataseconjugated avidin and substrate. Decreasing concentrations of biotinylated heparin bound ACP in the presence of increasing concentrations of unlabeled competitor (Fig. 3). Specifically, 1000 g/ml unlabeled heparin inhibited binding of labeled heparin by 84.1%. These results suggest that ACP binds specifically to heparin.
Soluble GAG Derived from ME180 Cells Inhibits Interaction of ACP with ME180 Cells-To assess the interaction between ACP and cell surface GAG more directly, we purified ME180 cell surface GAG from other cellular components using the methods described above. The resulting GAG was added, in varying concentrations, to labeled ACP and incubated at 4°C for 1 h before this mixture was added to confluent ME180 cells. The cells were subsequently analyzed by flow cytometry and showed concentration-dependent inhibition of ACP interaction with ME180 cells in the presence of ME180-derived GAG (Fig.  4). At ϳ2 g/ml, the highest tested concentration of GAG, ACP staining was inhibited by 57.5%. Although we cannot eliminate the possibility that contaminants present in small amounts in our preparation of purified GAG are responsible for the observed effects, these data further support the hypothesis that full ACP binds to cellular GAG.
Kinetics of ACP Association with ME180 Cells-Based on our hypothesis that ACP recognizes ME180 surface GAG, we anticipated that soluble protein should associate with cells rapidly, within minutes. To test this hypothesis, we incubated confluent ME180 cells with fluorescently labeled protein (BSA or soluble ACP) at 37°C for varying periods of time up to 24 h. Soluble ACP stained 21.08% of cells at 30 min, 64.15% of cells at 60 min, 95.09% of cells at 4 h, 99.22% of cells at 12 h, and 99.67% of cells at 24 h (Fig. 5). Of note, the majority of signal accumulated after 30 min of incubation. The mean levels of fluorescent signal per cell (data not shown) continued to in-crease during the later period as well, from 87.58 at 12 h, to 111.36 at 24 h. The BSA control samples changed minimally in signal over the entire time course. The gradual kinetics of cellular ACP accumulation, in combination with prior data indicating that ACP mediates internalization of GBS (5), led us to test the hypothesis that the soluble ACP might enter cells over time.
Soluble ACP Enters ME180 Cell Cytoplasm-We incubated confluent ME180 cells on a glass coverslip with labeled ACP for 4 h at 37°C. The cells were then washed, fixed, stained with mouse monoclonal EGFR-specific antibody and a rhodaminelabeled anti-mouse secondary antibody to mark the cell surface, and viewed by confocal microscopy. As shown (Fig. 6), ACP (green) is distributed throughout the cell interior, whereas EGFR (red) stains circumferentially. Images collected as a Z plane series (eight slices from top to bottom of the monolayer) include green signal in multiple planes, including slices 4 and 5, situated midway between the cell base and cell apex. Addi- FIG. 3. Alpha C protein binds specifically to soluble heparin. A binding inhibition assay, modeled on enzyme-linked immunosorbent assay techniques, was performed, in which wells were coated with alpha C protein (ACP), then incubated with 200 g/ml heparin-albumin-biotin and unlabeled heparin sodium in varying concentrations. Bound biotinylated material was detected with alkaline phosphataseconjugated avidin and substrate. Soluble heparin sodium inhibited binding of heparin-albumin-biotin to ACP in a concentration-dependent manner, suggesting that this interaction is specific.

FIG. 4. Soluble glycosaminoglycan material derived from ME180 cells inhibits association of the cells with soluble alpha C protein in a concentration-dependent manner.
Glycosaminoglycan (GAG) was purified from 18 plates of ME180 cells using the methods described. The final precipitated material was resuspended in phosphate-buffered saline and incubated in varying concentrations with labeled alpha C protein (ACP) for 1 h at 4°C; the mixtures were then added to a confluent monolayer of ME180 cells and incubated at 37°C for 1.5 h. Cells were then washed, fixed, and analyzed by flow cytometry. Data are shown as percentage inhibition of staining (uninhibited sample % staining Ϫ inhibited sample % staining/uninhibited sample % staining) and reveal concentration-dependent inhibition of accumulation of cell-associated ACP.
tional studies (not shown) were performed using fluorescently labeled phalloidin to identify the actin cytoskeleton. Images from these studies also show internal signal from labeled ACP.
To further detect intracellular ACP, we evaluated the effect of an anti-AlexaFluor antibody (Molecular Probes) that quenches the fluorescence of AlexaFluor-488. This antibody (rabbit IgG) was added to ME180 cells that had been incubated with AlexaFluor-488-labeled ACP as described above. A rhodamine-labeled anti-rabbit secondary antibody was then added to detect the anti-AlexaFluor antibody. Because we used no cell-permeabilizing treatment, we expected that surface-bound ACP (green) signal would be quenched and replaced by rhodamine (red) signal, whereas internal ACP signal would persist (green), because this ACP is inaccessible to the anti-AlexaFluor and secondary antibodies. The images (not shown) reveal green signal surrounded by peripheral red signal. Control samples without the quenching antibody show diffuse green signal and minimal red background. Control samples without ACP also show minimal red background, along with minimal green signal. These data support our hypothesis that free ACP enters ME180 cells.
To confirm this finding and to determine the intracellular location of ACP, we incubated cells with ACP, washed away unbound material, and fractionated the samples either chemically (by selectively solubilizing the plasma membranes with 1% digitonin) or mechanically (by passing the cells through a 27-gauge needle), as detailed under "Experimental Procedures." Resulting proteins in cellular fractions were subjected to SDS-PAGE and evaluated by Western blot using a monoclonal antibody recognizing ACP. These data (Fig. 7a) show full-length ACP in both the digitonin-derived pellet (membranous/cellular material) fraction and the supernatant (cytosol) fraction. Negative control studies included ME180 cells incubated with biotinylated BSA and probed with avidin. As an additional control study, ME180 cells were incubated with beta C protein (a non-Alp family GBS surface protein), processed similarly, and probed with beta C protein-specific monoclonal antibody. Neither BSA nor beta C protein was detected in either supernatant or pellet after digitonin treatment. These data support the hypothesis that, in contrast to ACP, BSA and beta C protein do not bind or enter ME180 cells. Mechanical cellular fractionation samples (Fig. 7b) show ACP in the final supernatant, representing the eukaryotic cell cytosol.
To determine the mode of ACP entry into and trafficking within ME180 cells, we incubated cells under conditions expected to disrupt known mechanisms of receptor mobilization and ligand internalization. We chose agents to inhibit specific cellular processes, including actin polymerization (cytochalasin D), clathrin lattice formation (hypertonic sucrose and chlorpromazine), and transport of proteins from the endoplasmic reticulum to the Golgi complex (brefeldin A), then assessed for ACP binding and entry into the treated cells.
Cytochalasin D Treatment Inhibits Interaction of ACP with ME180 Cells-Confluent ME180 cells were incubated with cytochalasin D (2.5 g/ml), which inhibits de novo actin polymerization, at 37°C for 30 min prior to addition of fluorescently labeled protein (BSA, soluble ACP), incubation at 37°C, and processing for analysis by flow cytometry or confocal microscopy. Results were compared with those for cells processed identically but without cytochalasin D treatment. ACP association with cells was markedly diminished by cytochalasin D. Specifically, the percentage of cells staining with ACP by flow cytometry (Fig. 8A) decreased from 72.34% to 12.45% with cytochalasin D treatment. BSA samples stained minimally regardless of cytochalasin D treatment, and CD46 and EGFR expression were quantitatively unaffected by cytochalasin D treatment. Representative confocal microscopy images (Fig.  8B) show markedly less staining of the cytochalasin D-treated cells than of the untreated samples for ACP (green) and an alteration in the staining pattern for EGFR (red). These data indicate that ACP association with ME180 cells requires actin polymerization. They are consistent with other work showing that EGFR interacts with actin (25,26).
ACP accumulation was also diminished by inhibitors of clathrin-dependent endocytosis. Specifically, pretreatment of ME180 cells with hypertonic sucrose (0.3 M) or with chlorpromazine (25 g/ml) reduced cellular ACP association by 92.4% and 61.8%, respectively, without influencing the staining of EGFR or CD46 (data not shown). These data demonstrate that interaction of ACP with ME180 cells requires clathrin-dependent endocytosis.
No change in ACP staining was noted by flow cytometry for cells treated with brefeldin A (5 M), which blocks protein transport out of the endoplasmic reticulum. These data indicate that the vesicle transport pathways from endoplasmic reticulum to Golgi are not essential for ACP interactions with ME180 cells.
C. difficile Toxin B Inhibits Association of Alps with ME180 Cells-The results of the cytochalasin D studies described above indicate that interactions of ACP with ME180 cells depend upon actin polymerization. Because the Rho GTPases Rho, Rac, and Cdc42 are known to regulate pathways that control cellular actin rearrangements, we hypothesized that one or more of these molecules might play a role in the actindependent association of ACP with ME180 cells. To test this hypothesis, we pretreated confluent ME180 cells with C. difficile toxin B (List Biological Laboratories), an exoenzyme that selectively inactivates the Rho subfamily of GTPases by glucosylating the threonine at position 37 in Rho or at the corresponding position in Rac and Cdc42, using UDP-glucose as a co-substrate (27,28). We incubated the cells with varying concentrations of toxin for 24 h prior to adding fluorescently la-beled BSA or soluble ACP and assessing by flow cytometry. In comparison to untreated cells, the toxin-treated cells interacted less effectively with ACP; toxin treatment inhibited ACP association with ME180 cells by as much as 84.6% (Fig. 9).
N-terminal and Repeat Region Domains Are Necessary for Full ACP Activity-To identify the region(s) of ACP binding cellular GAG to allow entry into ME180 cells, we labeled soluble 1-repeat ACP derived by recombinant methods and tested this protein for interaction with ME180 cells, analyzing by flow cytometry as described above. In addition, we tested separate protein constructs representing the N-terminal region of ACP

FIG. 8. Cytochalasin D diminishes association of alpha C protein with ME180 cells.
A, confluent cells were incubated with cytochalasin D at 2.5 g/ml for 30 min, then with directly labeled soluble alpha C protein (ACP) or bovine serum albumin (BSA). Cells were then washed, fixed overnight, and analyzed by flow cytometry. CD46-specific antibody and a fluorescent secondary antibody were used in control samples. Compared with untreated samples, cytochalasin D-treated samples stained poorly with soluble ACP. Epidermal growth factor receptor (EGFR) and CD46 antibody staining, and BSA staining (not shown) were unchanged by cytochalasin D treatment. B, confluent ME180 cells were incubated with labeled soluble ACP (green), then washed, fixed, and stained with EGFR-specific antibody and rhodamine-labeled secondary antibody (red) to mark the cell surface. Samples were untreated (top) or treated with 2.5 g/ml cytochalasin D for 30 min (bottom) prior to incubation with labeled protein.
Cytochalasin D-treated cells showed markedly lower signal from ACP and an altered pattern of staining with EGFR antibody. and the 9-repeat region (9RR) of ACP; these were derived previously using recombinant techniques (5,15). Flow cytometry analysis showed that a full-length 1-repeat ACP construct associated with ME180 cells similarly to full-length 9-repeat ACP (86% of cells stained after 1.5-h incubation with 0.3 M labeled protein). However, the N-terminal and 9RR proteins associated less effectively than full-length ACPs, staining 1.75% and 2.19% of cells (respectively) after 1.5-h incubation with labeled protein. Furthermore, addition of both the Nterminal and 9RR protein constructs to ME180 cells simultaneously did not alter the degree of staining. These data demonstrate that the N-terminal and 9RR domains must be contiguous to interact optimally with ME180 cells. Consistent with the observed lack of binding, neither of these truncated ACP derivatives was found to be internalized in cellular fractionation assays.
To evaluate for heparin-binding activity in the ACP domains, we performed dot blot analysis of recombinantly derived ACP products (full-length 1-repeat ACP, 9-repeat ACP, N-terminal of ACP, and 9RR of ACP); BSA was included as a negative control protein (Fig. 10). Proteins were dotted on nitrocellulose membrane, incubated with heparin-albumin-biotin (Sigma, 0.05 mg/ml), washed, and detected with alkaline phosphataseconjugated avidin/substrate. Data (Fig. 11) show binding of heparin to full-length 1-repeat and 9-repeat ACP, with less binding to 9RR, and no apparent binding to the ACP N-terminal region alone. Based upon these data, we suggest that both the N-terminal domain and the repeat region domains contribute to the GAG-binding and cell entry properties of ACP.
Soluble Alp3 Interacts with ME180 Cells Similarly to ACP-To determine whether the highly homologous DNA and amino acid sequences among Alp family members correlate with similar functional properties, we produced soluble Alp3 protein by amplifying, cloning, and expressing the gene Alp3 from GBS type V strain H4A0158. As we have shown previously for soluble ACP, incubation of varying concentrations of soluble Alp3 with ME180 cells for 1 h prior to adding live GBS strain A909 (expressing ACP) inhibited bacterial entry (Fig.  12). As expected, Alp3 inhibited internalization with less potency than soluble ACP; specifically, 3 M Alp3 inhibited A909 entry of ME180 cells by 78.6%, whereas the same concentration of soluble ACP inhibited entry by 87.9%. These data demonstrate that ACP and Alp3 function similarly.
Additional studies, analogous to those we have described for ACP, support this hypothesis. Specifically, flow cytometry analysis showed comparable levels of ME180 cell staining with soluble Alp3, following kinetics similar to those described for soluble ACP. Pretreatment of ME180 cells with cytochalasin D or C. difficile toxin B diminished association with soluble Alp3, similar to the results seen for ACP. Cellular fractionation after ME180 cells were incubated with soluble Alp3 revealed cytoplasmic Alp3, analogous to the results of similar assays using soluble ACP. Finally, as described for ACP, soluble Alp3 interaction with ME180 cells was inhibited by soluble heparin sodium in a concentration-dependent manner, and soluble Alp3 bound to heparin-albumin-biotin in a dot blot assay. In total, these data demonstrate that ACP and Alp3 function similarly and suggest that other Alps might share these functions also. DISCUSSION GBS frequently colonize mucosal surfaces but infrequently invade deeper tissues. The molecular events underlying colonization and invasion remain poorly understood. Although GBS entry into epithelial cells and transcytosis across layers of epithelial cells have been described in the literature (5), the route of transcytosis is unknown. Specifically, GBS may traverse the cell layers by an intracellular route, or by a paracellular route (as has been described for GAS (29)), or both. Thus, it is unclear whether GBS entry into epithelial cells represents an intermediate step in transcytosis and thereby facilitates invasive disease, as has been described for numerous intracellular pathogens (Mycobacterium tuberculosis, Shigella species, Salmonella species, and others), or whether GBS entry into epithelial cells represents a mechanism of host defense, in that epithelial cells containing internalized bacteria may be shed and eliminated from the host, as has been described for Pseudomonas aeruginosa (30). It is also possible that bacterial internalization serves as both a GBS virulence mechanism and a host defense mechanism. The attenuated virulence of a GBS strain lacking ACP (4) and the association of ACP with GBS internalization of epithelial cells in vitro (5) support an association of GBS internalization of epithelial cells with invasive illness. In this context, the interaction of ACP with epithelial cells may represent a key step in the transition from colonization to active infection. In this work, we have identified host cell GAG as a receptor for ACP and have shown that ACP is internalized into host cells by a mechanism requiring Rho Full-length alpha C proteins (ACPs) include N-terminal (white), C-terminal (black), and repeat regions (gray). For the dot blot assay, full-length ACPs bearing 9 repeats or a single 82-amino acid "repeat" region as well as constructs corresponding to the N-terminal region of ACP and the 9-repeat region of ACP were studied.
GTPase-mediated actin polymerization. Furthermore, we have found that both the N-terminal and repeat region domains of ACP are important for these events.
The data implicating host GAG structures, and in particular heparan sulfate proteoglycans (HSPGs), as receptors for ACP include: 1) inhibition of this interaction by pretreatment of ME180 cells with sodium chlorate, which prevents sulfation of eukaryotic cell structures (24); 2) inhibition of this interaction by pretreatment of ME180 cells with heparitinases but not chondroitinase ABC; 3) inhibition of this interaction in the presence of soluble heparin or heparan sulfate (but less so by chondroitin or other GAG molecules tested); 4) binding of ACP to heparin-albumin-biotin in vitro; and 5) inhibition of this interaction by GAG derived from ME180 cells. Further support for our hypothesis that ACP binds to a heparin-like eukaryotic receptor comes from crystallographic structural studies of the N-terminal region of ACP. These studies show regions of basic residues similar to heparin binding domains in other proteins; one such site begins in the N-terminal region and extends presumably into the repeat region as well. 2 Found on eukaryotic cell surfaces and in the extracellular matrix, proteoglycan molecules include chains of sulfated polysaccharide GAGs anchored in the plasma membrane by a core protein that may associate with the actin cytoskeleton intracellularly. Proteoglycans participate in diverse processes, including cell adhesion, growth factor signaling, tumor biology, hemostasis, and lipid metabolism. These molecules also medi-ate interaction between epithelial cells and numerous pathogens, including viruses (such as human immunodeficiency virus (10), herpes simplex virus (31), human papilloma virus (9), and parainfluenza 3 (7)), spirochetes (such as Borrelia burgdorferi (32)), parasites (such as Trypanosoma cruzi (8)), and bacteria (such as Neisseria gonorrhoeae (11) and Listeria monocytogenes (6)). In some cases, GAG serves as a co-receptor, modulating the interaction of an extracellular ligand with other host cell surface structures by altering ligand concentration, stability, or conformation, or by promoting ligand or receptor oligomerization (33). In particular, HSPGs may facilitate interactions between ligands and cell-surface integrins (34 -36).
The described cell surface proteoglycans include syndecans (four members), glypicans (six members), betaglycans, and others. Extensive work by others has shown that HSPG expression is cell-specific, tissue-specific, and developmental stage-specific (37). Most cells express at least one syndecan, and many express multiple syndecans; for example, syndecan-1 is found on epithelial cells and malignant plasma cells, syndecan-2 is found on fibroblasts, and syndecan-3 is found predominantly in the central nervous system, whereas syndecan-4 is ubiquitous (38). Our finding that the mixture of free chondroitin sulfates A and C partially inhibits the interaction of ACP with ME180 cells suggests that the molecule recognized by ACP may be a syndecan, because these molecules contain both heparan and chondroitin-like moieties. Glypicans are also widely expressed and are found commonly in addition to syndecans on the cell surface. Notably, HSPG expression may vary with physiological and pathophysiological states, such as pregnancy, malignancy, diabetes, and chronic ulcer (39 -42). The importance of HSPGs for Alp recognition, in combination with the variable profile of HSPG expression among different cell types, tissue types, and stages of development, may explain the tissue tropism and variability of risk for GBS colonization and infection.
Our data are consistent with the hypothesis that binding of ACP to HSPGs leads to cellular uptake of ACP, mainly via clathrin-dependent endocytosis. Because cytochalasin D treatment or C. difficile toxin B treatment of ME180 cells diminishes cellular ACP accumulation, including surface staining in confocal images, we believe that actin polymerization is required for optimal ACP-receptor interaction at the cell surface. Apparently induced by the presence of free ACP, the actin-dependent process may include clustering, recruitment, recycling, affinity modulation, or synthesis of receptors; these processes require actin polymerization in other systems (43)(44)(45)(46)(47)(48). The tempo of the interaction and the lack of effect of brefeldin A treatment on association of ACP with ME180 cells suggest that de novo protein synthesis is not required for the interaction. Consistent with the data implicating HSPGs as an ACP receptor that facilitates actin-mediated downstream events, others have reported that HSPGs interact with the actin cytoskeleton, and in particular that ligand binding induces HSPG aggregation along actin filaments (49,50).
The subsequent events in ACP entry into the cell and traf-2 T. C. Aupérin and G. R. Bolduc, personal communication.
FIG. 11. Alpha C protein binds heparin by dot blot. Full-length 1-repeat alpha C protein (ACP) and 9-repeat ACP, as well as the N-terminal region ACP protein, 9RR, Alp3, and BSA proteins were immobilized on nitrocellulose in equal quantities and then incubated with heparin-albumin biotin, washed, incubated with avidin, washed, and developed. As shown, 1 g of full-length ACP proteins bound heparin more effectively than the same amount of either the N-terminal protein or the 9RR protein. Full-length Alp3 protein bound heparin also. In blots using less protein per dot, visible signal from heparin binding remained when the amount of full-length 9-repeat ACP was diminished to as low as 0.04 g.

FIG. 12. Soluble Alp3 inhibits internalization of group B
Streptococcus strain A909. Alpha C protein (ACP) or soluble Alp3 (alphalike protein 3) was added, in varying concentrations, to an assay in which group B Streptococcus strain A909, expressing ACP, was incubated with ME180 cervical epithelial cells. Cells were then incubated with penicillin and gentamicin to kill external bacteria, detached with trypsin-EDTA, and lysed with Triton X-100. Lysates were plated to allow enumeration of bacterial colony-forming units. Maximal internalization (100%) is based on the amount of internalized A909 in the absence of inhibitor. Soluble ACP and Alp3 proteins both inhibited internalization of A909. Control experiments (data not shown) included addition of bovine serum albumin (BSA) at comparable concentrations, which did not inhibit bacterial internalization.
ficking to the cytoplasm may also depend on actin rearrangements. In support of this hypothesis, reports in the literature describe actin-dependent GBS entry of epithelial cells (51)(52)(53) as well as actin-dependent uptake activity induced in epithelial cells by ligand binding to surface HSPGs (50). Actin rearrangements facilitate entry of other bacterial species into eukaryotic cells as well, including Yersinia, Salmonella, and Shigella species, and L. monocytogenes.
The inhibitory effect of C. difficile toxin B on ACP association with ME180 cells suggests the Rho GTPases mediate actin rearrangements required for this interaction to occur. The Rho GTPases Rho, Rac, and Cdc42 are molecular switches that regulate cellular actin rearrangements through three distinct pathways to mediate changes in cell shape, contractility, adhesion, migration, gene transcription, cytokinesis, membrane trafficking, and growth (54). These GTPases are important targets in the pathogenesis of numerous organisms, including C. difficile, Salmonella species, Shigella species, and pathogenic E. coli. C. difficile promotes infection through toxins that interfere with Rho GTPases signaling. Salmonella, Shigella, and certain pathogenic E. coli induce their own internalization or tight adherence to the cell surface by triggering actin rearrangements through the modulation of Rho GTPase activity (55)(56)(57)(58)(59)(60)(61)(62). The role of Rho GTPases in the pathogenesis of Grampositive infections is less clear. GAS entry into eukaryotic cells involves Rac and Cdc42 (63). Our data indicate that members of the Rho GTPase family mediate the cellular events required for binding and entry of ACP into ME180 cells. Identification of the molecular links between ACP and a Rho GTPase-mediated pathway, and of the specific downstream elements within the pathway(s), remains the subject of further study.
Integrins warrant consideration as participants in this process. These cell surface molecules interact with Rho GTPases (64) and have been recognized both as HSPG co-receptors (as noted above) and as receptors for microorganisms (65). Further data supporting possible integrin binding to ACP comes from crystallographic structural studies of the N-terminal region of ACP that reveal similarities to the structure of the integrinbinding protein fibronectin, including a potential integrin binding motif. 2 Thus, both the mechanism and the pathogenic significance of intracellular ACP trafficking to the cell cytosol are unknown. The effects of ACP in mediating GBS entry into ME180 cells (5) suggest that soluble ACP may enter cells by the same mechanism as whole GBS expressing ACP. Although GBS has not traditionally been considered an intracellular pathogen, many researchers have described internalization of the organism; most internalized GBS appear in vacuoles (51, 66 -68), but some appear in the cytoplasm (69). Our data are consistent with the hypothesis that cytoplasmic ACP may derive from cytoplasmic GBS. Alternatively, ACP may detach from the surface of GBS either before or after GBS enter host cells. In either case, we believe the role of ACP in facilitating GBS transcytosis is likely to be mediated by its interaction with a cell surface GAG and perhaps by associated changes in the actin cytoskeleton. In other systems, GAG-binding ligands may cause changes in junctions between host cells that make tissues vulnerable to paracellular penetration. For example, the GAG-binding growth factor VEGF can transiently increase endothelial permeability (70), and heparin-binding hemagglutinin adhesin, a GAG-binding protein from M. tuberculosis, contributes to dissemination of disease to extrapulmonary sites (71). Future work will be required to identify potential non-GAG co-receptors for ACP and to study the interactions of the relevant host surface molecules with whole GBS.
The structural similarities among Alp family proteins, including their homologous N-terminal regions and their variable numbers of tandem repeats, suggest that these proteins might function similarly to ACP in pathogenesis. Specifically, the N-terminal region of Alp3 is 77.9% identical to that of ACP, whereas the repeat regions of these proteins are 38% identical. Our data indicate that soluble Alp3 and ACP associate with ME180 cells by an actin-dependent, Rho GTPase-dependent mechanism, on a similar time course, and with similar degrees of inhibition by free heparin. In addition, Alp3 inhibits entry of ACP-expressing GBS strain A909 into ME180 cells. These data demonstrate that in fact these Alps share similar properties, which might reasonably be attributed to their highly homologous N-terminal regions. The ability of 1-repeat ACP to interact with ME180 cells also supports the hypothesis that the N-terminal region plays an important role. However, although the N-terminal region of ACP effectively inhibits ACP-mediated internalization of whole GBS (5), this region does not effectively bind or enter ME180 cells on its own, and, in comparison to full-length ACP, it binds minimally to heparin-albumin-biotin. The 9RR construct also fails to bind or enter ME180 cells effectively and binds less effectively than full-length ACP to heparin-albumin-biotin in vitro. Incubation of ME180 cells simultaneously with soluble N-terminal ACP and 9RR proteins yields data similar to those seen for each of these components alone, suggesting that both N-terminal and repeat regions must be present in the same molecule to interact with ME180 cells effectively. Although it is possible that the domains fold improperly when expressed separately, these data are also consistent with a hypothesis that a junctional epitope between the two domains may be important for the interactions. Supporting this hypothesis are the structural data described above, revealing a putative heparin-binding domain in the distal portion of the N-terminal of ACP that extends (presumably) into the repeat region (Fig. 13). Alternatively, it is possible that binding of each of two domains is required, one within the N-terminal region and another within the repeats, and the two binding domains may need to be in the correct relative configuration for binding to be stable and for protein internalization to proceed.
In summary, our data indicate that soluble ACP interacts with host GAG, then enters the ME180 cell cytosol by an actin-dependent, Rho GTPase-mediated mechanism. These properties may be shared among Alps and seem to involve both FIG. 13. Schematic of 9-repeat ACP structure. ACP includes an N-terminal domain, repeat region(s) of 82 amino acids each, and Cterminal domain. Numbers to the right of the diagram refer to amino acid number; the first 56 amino acids encode a signal sequence that is cleaved prior to surface expression. Structural analysis indicates that residues in the distal portion of the N-terminal region (Asp 160 -Leu 226 ) form a pocket that is compatible with heparin-binding activity. 2 Residues in the adjacent portion of the repeat region may also contribute features compatible with heparin-binding activity.
N-terminal and repeat-region domains. Further studies are warranted to further characterize the host cell HSPG and a possible co-receptor, determine the route by which ACP reaches the cytosol, and delineate the subsequent Alp-associated events in GBS pathogenesis.