Fig 1.
Bsp Distribution and Conservation.
(A) The amino acid sequences listed in S1B Fig were used to form a neighbor-joining phylogenetic tree calculated by a Blosum80 cost matrix and Jukes-Cantor genetic distance model. The scale bar in the center indicates the distance representing 0.03 substitutions per site. Serotype of each strain is indicated by a colored node, while sequence type is indicated by a number within each node. This tree is overlaid on a pie chart which shows the distribution of each bsp homolog (also shown in the bottom right). (B) The gene structure for COH1 bspC is shown, with specific domains and relevant regions within the V-domain annotated.
Fig 2.
Mapping vimentin binding activity to the BspC V-domain.
(A) The V-domain of BspC was modeled using BspA as a template. Surface residues are shown with hydrophobic residues colored in orange (top) and electrostatic potential as calculated by APBS (bottom) colored with red indicating higher potential and blue indicating lower potential. APBS calculation and visualization were performed using PyMOL. (B) Circular dichroism spectrum of a 1 mg/mL solution of the BspC V-domain. (C) The V-domain was added to 20 nM vimentin at the indicated concentrations prior to measuring the microscale thermophoresis dose response curve quantifying the dissociation constant. (D) hCMECs were pretreated with PBS (vehicle), the V-domain added to a concentration of 10 μM, or 10 μM of denatured V-domain 30 minutes prior to infection. CFU were plated and to assess V-domain blocking of GBS adherence after 30 minutes of incubation. B-D display means from 3 independent experiments. Error bars indicate standard error of the mean. Statistical analysis: (D) One-way ANOVA with Tukey’s multiple comparisons. ***, P < 0.0005.
Fig 3.
Characterization of the role of the V-domain gating loop.
(A) Circular dichroism spectrum of a 1 mg/mL solution of the BspC GLM V-domain. (B) The GLM V-domain was added to 20 nM vimentin at the indicated concentrations prior to measuring the microscale thermophoresis dose response curve quantifying the dissociation constant. (C) hCMECs were pretreated with PBS (vehicle), the V-domain added to a concentration of 10 μM, or 10 μM of the GLM V-domain 30 minutes prior to infection. CFU were plated to assess V-domain blocking of GBS adherence after 30 minutes of incubation. (D) COH1ΔbspC GBS containing pDCErmbspC (pbspC), pDCErmbspC GLM (pGLM), or the empty vector (pDCErm) were stained with an anti-BspC antibody or isotype control without permeabilization. Surface bound antibody was measured via flow cytometry. (E) The same strains used in (D) were used to infect a monolayer of hCMECs. CFU were plated to assess adherence after 30 minutes of incubation. A-C and E display means pooled data from 3 independent experiments. D displays representative histograms taken from one of three independent flow experiments. Error bars indicate standard error of the mean. Statistical analysis: Two-way ANOVA with Tukey’s multiple comparisons (C) and One-way ANOVA with Tukey’s multiple comparisons (E). **, P < 0.005; ****, P < 0.00005.
Fig 4.
Flexibility of the Pocket-Guarding Gating Loop is Required for GBS Meningitis.
Mice were infected with ~4x108 CFU of COH1ΔbspC GBS containing pDCErmbspC (pbspC), the empty vector (pDCErm), or pDCErmbspC GLM (pGLM) and then sacrificed after 48 hours. (A) Half of each mouse brain was homogenized and plated on THA-Erm to determine bacterial burden. (B) Blood from each mouse was also plated on THA-Erm to determine bacterial burden. For (A) and (B), data is pooled from three independent experiments, where each dot represents an individual mouse, and lines indicate statistical means. (C) H&E stains showing representative images of the leptomeninges from two individual mice per group. Arrows indicate areas of leukocyte infiltration and meningeal thickening. (D) ImageJ was used to analyze meningeal thickness from 3 mice per group. Two images were taken of each brain, and three distinct regions per image were measured with each dot representing an individual region. (E) KC protein concentration was quantified via ELISA from the brain homogenates shown in A. Data for E is pooled from three independent experiments, where each dot represents an individual mouse brain, and lines indicate statistical means. Statistical analysis: (A, B, C, and D) One-way ANOVA with Tukey’s multiple comparisons. *, P < 0.05; **, P < 0.005; ****, P < 0.00005.
Fig 5.
The V-domain pocket contributes to BspC Function (A) Zoomed in view of the V-domain pocket with residues targeted for mutagenesis. Residues shown in blue were targeted to construct LPM, and residues shown in green were targeted to construct RPM. Visualization done using PyMOL. (B) COH1 ΔbspC GBS containing pDCErmbspC (pbspC), pDCErmbspC LPM (pLPM), pDCErmbspC RPM (pRPM) or the empty vector (pDCErm) were stained with an anti-BspC antibody or isotype control without permeabilization. Surface bound antibody was measured via flow cytometry. (C) The same strains used in (B) were used to infect a monolayer of hCMECs. CFU were plated to assess adherence after 30 minutes of incubation. (D-H) Mice were infected with ~4x108 CFU of the same strains used in (B) and then sacrificed after 48 hours. (D) Half of each mouse brain was homogenized and plated on THA-Erm to determine bacterial burden. (E) Blood from each mouse was also plated on THA-Erm to determine bacterial burden. For (D) and (E), data is pooled from two independent experiments, each dot represents an individual mouse, and lines indicate statistical means. (F) H&E stains showing the leptomeninges from two individual mice per group, one from each independent experiment. Arrows indicate areas of leukocyte infiltration and meningeal thickening. (G) ImageJ was used to analyze meningeal thickness from 5 mice per group from the two independent experiments. Two images were taken of each brain, and three points per image were measured with each dot representing an individual point. (H) KC protein concentration was quantified via ELISA from brain homogenates shown in D. Data for H is pooled from two independent experiments, where each dot represents an individual mouse brain, and lines indicate statistical means. Statistical analysis: (C, D, E, G, and H) One-way ANOVA with Tukey’s multiple comparisons. *, P < 0.05; **, P < 0.005; ***, P < 0.0005; ****, P < 0.00005.
Table 1.
Top 10 hits from the PLANTS virtual structure-based screen of e-Drug3D library of FDA approved drugs against the GBS BspC V-domain.
Fig 6.
Lapatinib and Carfilzomib bind the BspC V-domain and prevent adherence to hCMECs (A) GBS was pretreated with DMSO (vehicle), 10 μM of Lapatinib, Cobicistat, Venetoclax, or Carfilzomib, or 1 μM of Paliperidone Palmitate or Tafluprost 30 minutes prior to infection of hCMECs. CFU were plated to assess blocking of GBS adherence after 30 minutes of incubation. (B) A model of Lapatinib (top) and Carfilzomib (bottom) bound to the V-domain pocket. Visualization done using PyMOL. (C and D) GBS was pretreated with either DMSO, Lapatinib (C), or Carfilzomib (D) at the indicated concentrations 30 minutes prior to infection of hCMECs. CFU were plated to assess blocking of GBS adherence after 30 minutes of incubation. The dashed line indicates the mean adherence of WT GBS pretreated with DMSO, while the dotted line indicates the mean adherence of the ΔbspC mutant pretreated with DMSO or the drugs. Pooled data from three independent experiments is shown. Error bars indicate standard error of the mean. Statistical analysis: (A) One-way ANOVA with Tukey’s multiple comparisons. *, P < 0.05; **, P < 0.005; ****, P < 0.00005.
Fig 7.
Carfilzomib Prevents GBS Brain Entry in vivo.
(A) Schematic generated using BioRender displaying the experimental design and injection/infection timepoints for the comparison of Carfilzomib or vehicle (10% DMSO in sesame oil) treatments. Mice were infected with ~8.5x108 CFU of either WT or ΔbspC GBS. Four WT GBS infected vehicle treated mice and three WT infected carfilzomib treated mice were euthanized early based disease severity. All remaining mice were euthanized at 72 hours post infection, at which time the brain tissues (B) and blood (C) were harvested for enumeration of bacterial CFU. Pooled data from two independent experiments is shown. Error bars indicate standard error of the mean. Statistical analysis: (A) Two-way ANOVA with Tukey’s multiple comparisons. *, P < 0.05.
Fig 8.
Summary of Anti-BspC Treatment to Reduce the Pathogenesis of GBS Meningitis.
GBS utilizes BspC to interact with vimentin expressed by endothelial cells to adhere to the BBB endothelium and cause meningitis. Site-directed mutagenesis identified a vimentin-binding pocket contained within the BspC V-domain. This structure-function determination informed a virtual drug screen that yielded a list of drugs, two of which were confirmed to block BspC dependent adherence brain endothelial cells in vitro. Of these drugs, Carfilzomib was the most effective and was also able to prevent GBS entry into the brain in vivo. Figure generated using BioRender.