(a) Real-time RT–PCR analysis of chemokines IL-8, IL-6, CXCL1, CXCL2 and CCL20 in hBMEC 4 h post infection with WT GBS (black bars) or isogenic mutants Δsrr-1 (grey bars) and ΔpilA (white bars). Transcript levels were normalized to GAPDH and fold change was determined as described in Methods; statistical analysis was performed with two-way ANOVA (Bonferroni test). (b) IL-8 secretion by hBMEC on infection with GBS WT strains (black bars) and isogenic ΔpilA mutants (white bars). Concentrations of IL-8 in hBMEC supernatants collected 4 h post infection were measured using ELISA, two-way ANOVA (Bonferroni test); **P<0.005; ***P<0.001. (c) Transcript levels of IL-8 following treatment (4 h) with PilA–GST fusion proteins from different GBS serotypes as well as control GST protein (0.5 μM) and E. coli endotoxin (0.3 ng ml⁻¹). Cells were treated with the amount of endotoxin that were detected in PilA–GST fusion proteins and GST control protein preparations with a endotoxin detection kit, as described in Methods; statistical analysis with one-way ANOVA (Tukey's multiple comparison test). All experiments were performed three times in triplicate wells, and bars represent the standard deviation of the mean of one representative experiment, **P<0.005; ***P<0.001. (d–f) Binding of purified PilA–GST protein to hBMEC. Cells were treated with PilA–GST (d) or GST (f) and then probed with either Anti-PilA antiserum or Anti-GST Ab. Nucleus is stained with DAPI. Magnification of the boxed area in 'd' is shown in 'e'. Scale bar, 20 μm.

(a) In vivo neutrophil recruitment was assessed by measuring MPO activity in mice (CD1, male, n=8 per group) skin homogenates 4 h post infection either with WT GBS (black squares) or ΔpilA mutant (red squares). MPO assays were performed on mice skin homogenates after subcutaneous injection with 1×10⁶ CFU of either WT GBS or ΔpilA mutant strain. (b) Bacterial counts in skin homogenates were assessed by plating serial-fold dilutions on Todd-Hewitt broth agar plates. Experiments were performed twice, a representative experiment is shown. Bars represent mean MPO levels or bacterial cfu, statistical analaysis was performed using Student's t-test; **P<0.005; NS, non-significant. (c–h) Neutrophil recruitment in the CNS was assessed following direct intracranial injection in 5-day-old mice (n=4 per group) 8–10 h post inoculation with WT GBS or ΔpilA mutant. Representative images of brains from two independent experiments are shown for two mice (c–e) infected with WT (same mouse in c,d) and two mice (f–h) infected with ΔpilA mutant (same mouse in f,g). Brain sections stained with hematoxylin and eosin (H&E) (c,f) depict meningeal thickening during WT infection (c, arrows) and PMN infiltration by immunohistochemical detection with FITC–anti-Ly6G Ab (d,e,g,h). Scale bar, 20 μm.

(a) Binding of WT GBS and ΔpilA mutant to immobilized collagen was assessed by crystal violet staining. Stain released from attached bacteria was quantified by measuring absorbance at 595 nm. Student's t-test. (b) Dose-dependent binding of recombinant PilA protein to collagen I or BSA compared with GST control protein was measured by ELISA as described in Methods. Adherence (c) and invasion (d) of WT GBS (black bars) and ΔpilA mutant (white bars) in the presence increasing amount of collagen I. Bacteria were preincubated with indicated concentration of soluble collagen, washed extensively and assayed for the ability to adhere and invade the hBMEC by antibiotic protection assay. Data are reported as percent adherent or intracellular bacteria relative to the inuput inoculum. All experiments were performed three times in triplicate and bars represent means and standard deviations of one representative experiment. Statistical analysis was performed using Student's t-test (a) or one-way ANOVA with Tukey's multiple comparison test (all other experiments) (*P<0.05; **P<0.005; ***P<0.001. (e) Phase contrast images of PilA–GST, GST or BSA-coated latex beads binding to hBMEC in the presence of collagen or BSA. Cells were extensively washed with PBS to remove any unbound or loosely bound beads and visualized using a Zeiss Axiovert 200 inverted microscope (Carl Zeiss). Arrows indicate beads on hBMEC surface. Scale bar, 20 μm.

(a) Co-localization of PilA-coated latex beads with FAK visualized by confocal microscopy. PilA-coated beads are stained with Anti-PilA antiserum (green), FAK with Anti-FAK monoclonal Ab (red) and nucleus with To-Pro (blue). Scale bar, 100 μm. Inhibition of GBS internalization (b) and IL-8 secretion (c) by hBMEC following pretreatment with α₂β₁ integrin function blocking monoclonal Ab compared with isotype control Ab. GBS penetration (d) and IL-8 release (e) in β₁ integrin knockdown hBMEC cells using shRNA targeting β₁ integrin RNA (hBMEC Δβ₁) or a control shRNA (scramble shRNA). Bacterial invasion was quantified by antibiotic protection assay and IL-8 secretion was measured by ELISA. Black bars represent WT GBS and white bars represent ΔpilA mutant. All experiments were performed three times in triplicate and bars represent the standard deviation of the mean of one representative experiment. Statistical analysis was performed using one-way ANOVA (Tukey's multiple comparison test); *P<0.05; **P<0.005; ***P<0.001; NS, non-significant.

(a) Detection of phosphorylated forms of FAK (Tyr-397), Akt (Ser-473) and Erk1/2 (Tyr-204) as well as non-phosphorylated forms in hBMEC lysates at different stages of infection with WT GBS and ΔpilA mutant. Cells were infected for indicated time points, and cell lysates collected were probed with respective antibodies. Adherence (b), Invasion (c) and IL-8 secretion from (d) FRNK/hBMEC compared with FAK–WT/hBMEC on infection with WT GBS (black bars) or ΔpilA mutant (white bars). hBMECs were transfected with the C-terminally truncated FAK derivative FRNK (FAK-related non-kinase), which acts as a dominant negative form when overexpressed (FRNK/hBMEC). Bacterial invasion (e,g) and IL-8 secretion in hBMEC (f,h) on infection with WT GBS (black bars) or ΔpilA mutant (white bars) in presence of PI3K inhibitor LY294002 (e,f) or MEK1/2 inhibitor U0126 (g,h). All experiments were performed three times in triplicate and bars represent the standard deviation of the mean. Statistical analysis was performed using one-way ANOVA (Tukey's multiple comparison test); *P<0.05; ***P<0.001; ns: non-significant. KC levels (i) and MPO activity (j) in mice (n=8 per group) skin homogenates 4 h post subcutaneous injection with either WT GBS (black squares) or ΔpilA mutant (red squares). 2 h before bacterial injection mice were injected subcutaneously with 1 mg kg⁻¹ of either U0126, LY294002 or DMSO as a control. Experiments were performed twice and bars represent means and standard deviations of one representative experiment. All experiments were analysed by one-way ANOVA (Tukey's multiple comparison test); *P<0.05; **P<0.005; ***P<0.001; NS, non significant.

(a) Kaplan–Meier survival curve of mice following i.v. injection with WT GBS (black sqaures) or ΔpilA mutant (red squares). Groups of CD-1 male mice (n=10 per group) were injected intravenously with 7×10⁷ CFU of bacteria and survival was monitored over time and analysed using the Log-rank test; *P<0.05. Bacterial counts in the blood, spleen, brain (cfu per ml or cfu per g of tissue) (b) and ratio of bacterial counts in the brain and blood (c) 24 h post i.v. injection in mice (n=8 per group) with either WT GBS or ΔpilA mutant. (d) Transcript abundance of murine chemokine KC was determined by quantitative RT–PCR on messenger RNA samples isolated from mice brains, following infection with WT GBS or ΔpilA mutant for 24 h. Transcript levels were normalized to β-actin and expressed as fold change compared with mice injected with PBS only. Experiments were performed twice and bars represent standard deviation of the mean of one representative experiment. Statistical analysis was performed using Student's t-test; *P<0.05; NS, non significant. Histopathology of brain tissues of representative individual mice infected with WT GBS (e–h) compared with PBS treatment (i), or infected with ΔpilA mutant (j). Enlargement of boxed areas in 'e and g are shown in f and h' respectively, depicting meningeal thickening and neutrophil infiltration. (k–n). Immunohistochemical detection of GBS in the brains of mice infected with either WT GBS (k,m) or ΔpilA mutant (l,n). Bacteria were labelled with monoclonal anti-GBS Ab and detected with anti-mouse secondary antibody conjugated to HRP, coupled with hematoxylin counterstaining (k–l). GBS was also visualized by immunofluorescence with anti-mouse secondary Ab conjugated to AlexaFluor 488 (green). Blood vessels are stained with Anti-laminin Ab (red) and nuclei with To-Pro (blue) (m,n). Scale bar, 20 μm.

(a) Kaplan–Meier survival curve of mice following i.v. injection with 7×10⁷ CFU of WT GBS. 24 h before bacterial challenge groups of CD-1 male mice (n=10 per group) were treated intraperitoneally with 50 μg of either neutrophil-depleting Ab, RB6–8C5 (black sqaures) or isotype control Ab (red squares). Survival was monitored at least twice a day over a one-week period and data analysed using Log-rank test; P=0.07. Bacterial counts in the blood, brain (b) and ratio of bacterial counts in the brain versus the blood (c) 24 h post i.v. injection with WT GBS. (d) Pictures of mice brains following i.v. injection of EB dye 1 h before killing. (e) BBB permeability as reflected by amount of EB dye extravasation to the CNS. EB was extracted from brain tissue by treatment with 60% TCA and absorbance was measured at 610 nm. Vascular permeability of the EB dye is expressed as μg of EB per g of brain tissue using a standard curve. Experiments were performed twice and bars represent the standard deviation of the mean of one representative experiment. Statistical analysis was performed using Student's t-test; *P<0.05; **P<0.005; ***P<0.001; NS, non significant. (f–i) Permeability of the BBB as demonstrated by extravasation of FITC–Albumin (10 ml kg⁻¹) that was injected i.v. 30 min before killing. Brains from representative mice are shown following treatment with PBS only (f), infection with WT GBS following treatment with either RB6–8C5 Ab (g) or control IgG (h,i). Arrows indicate areas of dye leakage in the brain. Scale bar, 20 μm.

(1) PilA binds the extracellular matrix component collagen, which then engages α₂β₁ integrins on brain endothelium. (2) This leads to FAK activation and subsequent intracellular signalling. (3) Signalling pathway involving PI3K results in actin rearrangement and bacterial uptake. (4) A parallel signalling pathway involving MEK1/2 and Erk1/2 activation leads to IL-8 secretion. (5) This further results in functional neutrophil chemotaxis and activation. (6) Increased neutrophilic infiltrate damages the BBB resulting in increased BBB permeability, which likely facilitates further bacterial passage from the blood stream to the CNS.