An in vitro human vessel model to study Neisseria meningitidis colonization and vascular damages
- Institut Pasteur, Université Paris Cité, INSERM UMR1225, Pathogenesis of vascular infections, France
- Institut Pasteur, Université Paris Cité, Biomaterials and Microfluidics core facility, France
Figures
Figure 1 with 1 supplement
Replicating the geometries of infected in vivo vessels using photoablation.
(A) Confocal images of N. meningitidis-infected vessels in the human-skin xenograft mouse model. Scale bar: 25 µm. (B) Schematic representation of the development of the Vessel-on-Chip (VoC) device: (i) a collagen-based hydrogel is loaded in the center channel of the microfluidic device, (ii) the focused UV-laser locally carves the chosen geometry within the collagen I matrix, (iii) HUVECs are seeded and attach on the collagen-carved scaffold. (C) Schematic representation of the microfluidic device and zoom of the carved region after cell seeding (F-actin). Scale bar: 250 µm. (D) Confocal images of F-actin and the related orthogonal view of the tissue-engineering VoC. Scale bar: 50 µm. (E) Confocal images of VoC in circle-, square-, and diamond-shaped structures. Scale bar: 50 µm. (F) Photoablation process to create in vivo-like structures. Definition of the region of interest (ROI) to replicate from intravital microscopy images (left), and the obtained in vivo-like structures in the VoC (right). Scale bars: 20 µm.
Figure 1—figure supplement 1
Development, optimization, and characterization of a tissue-engineered Vessel-on-Chip platform based on photoablation.
(A) Schematic representation of the in vivo conditions assessed in this study: human vessels imaged in the human-skin xenograft. Scale bar: 30 µm. (B) Channel height depends on UV-Laser energy for 2.4 mg/ml (FujiFilm), 4 mg/ml (Corning), and 6 mg/ml (Corning) collagen I gels. (C) Orthogonal view of a bear and endothelial-layered collagen-carved tube. Scale bar: 20 µm. (D) Collagen-carved scaffold for 2.4 mg/ml (FujiFilm), 4 mg/ml (Corning), and 6 mg/ml (Corning) collagen I gel before cell seeding. Scale bar: 50 µm. (E) Confocal image of the VoC in 2.4 mg/ml - 3.5 mg/ml - 4 mg/ml and 6 mg/ml collagen I. Fujifilm collagen I has been used to make the 2.4 mg/ml (cyan circle) and Corning collagen I has been used to make the solution from 3.5 mg/ml to 6 mg/ml (blue circle). Scale bar: 30 µm. (F) Graph representing the number of sprouts per vessel for four collagen I concentrations: 2.4 mg/ml (FujiFilm), 3.5 mg/ml, 4 mg/ml, and 6 mg/ml (Corning). Each dot represents one vessel. For each condition, the mean ±s.d. is represented (2.4 mg/ml: 5.20 ± 6.0 (n=28) — 3.5 mg/ml: 1.69 ± 3.04 (n=29) — 4 mg/ml: 0.045 ± 0.22 (n=21) — 6 mg/ml: 0.23 ± 0.43 (n=30)) (G) Visco-elasticity measurements of collagen gels with a rheometer. Each dot represents a measure of each gel. For each condition, the mean ± s.d. is represented (FujiFilm, 2.4 mg/ml: (n=3) — Corning, 2.4 mg/ml: (n=3)–3.5 mg/ml: (n=3)–4 mg/ml: (n=3)–6 mg/ml: (n=3)). The mentions (Murray et al., 2022) and (Mairey et al., 2006) in (D), (F), and (G) refer to the providers of the collagen I, FujiFilm and Corning, respectively.
Figure 2 with 1 supplement
The Vessel-on-Chip device provides nuclear morphologies under flow conditions and permeability levels similar to those observed in vivo.
(A) Confocal images of the VE-Cadherin staining and (B) Collagen IV in the VoC. Scale bars: 50 µm (large view) and 7 µm (zoom). (C) Representative confocal images of fluorescent 150 kDa-Dextran (FITC) in the VoC (top, scale bar: 50 µm) and in the human vessel in the mouse model (bottom, scale bar: 30 µm). Fluorescence of the outside and inside regions of the vascular lumen has been measured to determine permeability. (D) Graph representing the permeability to 150 kDa-Dextran. Each dot represents one vessel. For each condition, the mean ± s.d. is represented (VoC, 2.4 mg/ml: 65.12 ± 26.00 cm/s (n=7) – VoC, 4 mg/ml: 3.97 ± 2.49 cm/s (n=5) - Human vessel in vivo: 8.00 ± 2.52 cm/s (n=7)). (E) Representative images of nucleus alignment in the absence and presence of flow (24 hr). Scale bar: 50 µm. (F–G) Graphs of nucleus orientation (IRQ: interquartile range) and elongation. Each dot corresponds to the mean value of one vessel. For each condition, the mean ± s.d. is represented (No flow: 53.0 ± 30.2º and 1.63 ± 0.11 a.u. (n=34)–2 hr of flow: 29.9 ± 11.5º and 1.85 ± 0.25 a.u. (n=22)–24 hr of flow: 18.2 ± 7.61º and 1.98 ± 0.20 a.u. (n=25) – in vivo: 12.9 ± 4.6º and 2.01 ± 0.23 a.u. (n=9)). All statistics have been computed with Wilcoxon tests.
Figure 2—figure supplement 1
Endothelial cells in the Vessel-on-Chip devices respond to flow-induced shear stress.
(A) Confocal images of PECAM-1 staining of the VoC. Scale bar: 50 µm. (B) Intravital images of PIV tracking of an arteriole and a venule in the mouse model (ear skin dermis). Scale bar: 50 µm. (C) Graph representing the flow rate depending on the vessel size. The red and blue fitting curves follow the equation for arterioles and venules, respectively. (D) In vivo wall shear stress of blood flow in arterioles and venules in the mouse model. Each dot represents a vessel. For each condition, the mean ± s.d. is represented (Arterioles: 1.57 ± 1.05 Pa (n=31) – Venules: 0.31 ± 0.35 Pa (n=12)). (E) Schematic of VoC experiments under flow conditions, controlled with a syringe pump. (F) Representative images of nucleus alignment (top) with the corresponding segmented/color-coded image (middle) and the circular plot of orientation (bottom). Scale bar: 30 µm. (G) Graph of nucleus interquartile range (orientation) and (H) nucleus elongation according to Wall Shear Stress (WSS). (I) Vessel diameter ratio before and after cell seeding (/) and (J) nearest nucleus distance for no flow condition, after 2 hr and 24 hr of flow. Each dot corresponds to a vessel. For each condition, the mean ± s.d. is represented (No flow conditions: 2.02 ± 0.74 and 14.90 ± 2.40 µm (n=34)–2 hr of flow: 1.42 ± 0.31 and 15.00 ± 1.52 µm (n=22)–24 hr of flow: 1.30 ± 0.27 and 14.9 ± 1.55 µm (n=26) — Animal model: 12.30 ± 2.70 µm (n=4)). All statistics have been computed with Wilcoxon tests.
Figure 3 with 1 supplement
Neisseria meningitidis adhesion to the Vessel-on-Chip device.
(A) Schematic of the infected Vessel-on-Chip device. (B) Confocal images of infected vessels in the VoC system 3 hr post-infection, in each condition. Top-left: z-projection of the infected VoC; top-right: slice of the middle of the infected VoC; bottom-left: 3D reconstruction of the infected VoC; and bottom-right: z-projection of the infected in vivo-like VoC. Scale bar: 20 µm. (C) Bright-field and fluorescence confocal images of the VoC 3 hr post-infection with WT (top) and pilD (bottom) Nm strains. Scale bar: 25 µm. (D) Graph representing the number of aggregates 3 hr post-infection, with wild-type (WT) and pilD Nm strains. 17 µm2 represents the median over the entire population of aggregate sizes. Each dot corresponds to a vessel. For each condition, the mean ± s.d. is represented (WT: 47.5 ± 30.4 (n=2) – pilD: 2.0 ±2.0 (n=3)). (E) Brightfield (left) and fluorescence (center) images of the infected vessels of different designs (circle, diamond, in vivo-like) and related simulation of shear stress. Scale bar: 50 µm. (F) Normalized histograms of the total number of pixels (blue dots and curve) and bacteria pixels (orange dots and curve) depending on the shear stress values for each vessel design (from top to bottom, circle, diamond, in vivo-like). Solid curves represent the mean ± sd (n=3 experiments). Spearman correlation gives the correlation between the mean curves.
Figure 3—figure supplement 1
Neisseria meningitidis uses T4P to adhere on the Vessel-on-Chip and its adhesion does not depend on the difference of shear stress.
(A–B) Graphs representing the numbers and the sizes of microcolonies within the VoC infected with wild-type (WT) or pilD Nm strains. For each condition, the mean ± s.d. is represented (WT: 189.0 ± 72.1 (n=2 vessels) and 16.60 ± 3.39 µm2 (n=378 microcolonies) – pilD: 14.3 ± 12.1 (n=3 vessels) and 9.03 ± 4.47 µm2 (n=43 microcolonies)). (C) Analysis pipeline of the correlation between bacterial location and shear stress in Vessel-on-Chip.
Figure 4 with 1 supplement
Flow impacts colony morphology but not growth in both Vessel-on-Chip and animal models, and N. meningitidis infection increases permeability in Vessel-on-Chip.
(A) Normalized surface area of microcolonies over time, in the VoC under flow conditions (left, n=20) and in human vessels of the xenografted mouse model (right, n=33). Solid thin curves, solid thick curves, and dashed thick curves represent individual colony growth, mean, and linear fit (y=ax + 1), respectively. (B) Colony doubling time extracted from the curves ( = 1/a). Each dot corresponds to a bacterial microcolony. For each condition, the mean ± s.d. is represented (VoC, Flow–: 16.7 ± 18.5 min (n=21) — VoC, Flow+ : 26.2 ± 40.4 min (n=20) — in vivo: 15.2 ± 18.8 min (n=33)). (C) Confocal images of wild-type (WT) Nm microcolonies formed on the vascular wall 3 hr post-infection, in the absence and presence of flow (VoC), and in the animal model. Scale bar: 10 µm. Circular plots representing the distribution of microcolony orientation. (D) Graph representing the elongation of microcolonies 3 hr post-infection in each condition. Each dot corresponds to a microcolony. For each condition, the mean ± s.d. is represented (VoC, Flow– : 1.75 ± 0.77 (n=128) — VoC, Flow+ : 2.11 ± 1.02 (n=125) — in vivo: 2.07 ± 0.971 (n=61)). (E) Brightfield and fluorescence confocal images of permeability assay (Dextran) in infected (top) and histamine-treated (bottom) Vessel-on-Chips. Scale bar: 25 µm. (F) Relative permeability values of VoC and in vivo vessels treated with histamine and infected with Nm. Each dot represents a vessel. For each condition, the mean ± s.d. is represented (mouse vessel, Control: 1 ± 0.62 (n=5), Histamine: 3.79 ± 2.25 (n=5) — VoC, Control: 1 ± 0.63 (n=5), Histamine: 5.11 ± 3.92 (n=7), Nm: 11.3 ± 5.48 (n=7)). All statistics have been computed with Wilcoxon tests.
Figure 4—figure supplement 1
N. meningitidis colonies grow and reorganize in vitro under flow conditions, and in vivo permeability is not affected by N. meningitidis at early stages of infection.
(A) Time-lapse images of bacterial growth in a Vessel-on-Chip (VoC) and an infected human vessel in the skin-xenografted mouse model. Scale bar: 30 µm. (B) Zoom on a bacterial fusion event in a VoC (top) and in a human skin-grafted mouse model (bottom). Scale bar: 10 µm. (C) Permeability of infected human vessels in the human skin-xenografted mouse model (green), normalized with the permeability of non-infected vessels (gray). For each condition, the mean ± s.d. is represented 70 kDa - Control: 1 ± 0.542 (n=11 vessels, N=3 mice), Nm: 1.34 ± 0.603 (n=11 vessels, N=3 mice) — 150 kDa - Control: 1 ± 0.314 (n=7 vessels, N=3 mice), Nm: 0.989 ± 0.280 (n=5 vessels, N=3 mice). Statistics have been done with Wilcoxon tests.
Figure 5 with 1 supplement
Flow-induced aligned actin stress fibers are reorganized below bacterial microcolonies.
(A) Confocal images of the F-actin network in the Vessel-on-Chip (VoC) in the absence and presence of flow (2 hr and 24 hr) (top). Scale bar: 30 µm. Corresponding segmented images. The color code shows the alignment of the actin fibers with the direction of the flow (red to blue) (middle). Circular plots of the orientation distribution of actin fibers (bottom). (B) Interquartile range of F-actin fiber orientation. Each dot represents the mean of F-actin fiber orientation per vessel. For each condition, the mean ± s.d. is represented (No Flow: 68.8º ± 31.0º (n=29) — 2 hr of flow: 33.1º ± 14.3º (n=25) — 24 hr of flow: 35.5º ± 13.5º (n=19)). (C) Confocal images of honeycomb-shaped cortical plaque formed by Nm microcolonies in the absence (top) and presence of flow (bottom) in the VoC. Scale bar: 10 µm. (D) 3D rendering of a vessel infected with Nm 3 hr post-infection under flow conditions. Scale bar: 15 µm (main) and 10 µm (zoom). (E) Percentages of colonies forming a cortical plaque in the 3D VoC (n=5 vessels) and 2D regions of the chips (n=4 lateral channels). (F) F-actin fluorescence intensity under each microcolony and on non-infected regions of the same area. Each dot represents a region of a microcolony. For each condition, the mean ± s.d. is represented (Not infected regions: 0.99 ± 0.33 a.u. (n=74) — Infection, without cortical plaque: 1.30 ± 0.57 a.u. (n=36) — Infection, with cortical plaque: 2.15 ± 0.54 a.u. (n=28)). (G) Coherency of F-actin fibers on non-infected regions and infection sites. Each dot represents a colony. For each condition, the mean ± s.d. is represented (Not infected regions: 0.40 ± 0.17 a.u. (n=74) — Infection, without cortical plaque: 0.28 ± 0.15 a.u. (n=36) — Infection, with cortical plaque: 0.18 ± 0.09 a.u. (n=28)). All statistics have been computed with Wilcoxon tests.
Figure 5—figure supplement 1
Flow-induced aligned actin stress fibers are reorganized below N. meningitidis microcolonies.
(A) Interquartile range of actin fibers according to the flow-induced wall shear stress. (B) Confocal images of F-actin cortical plaque at the infection sites in the 2D regions of the chips (n=5 chips, n=3 experiments). Scale bar = 40 µm. (C) Graph representing the correlation between F-actin mean fluorescence intensity and actin fiber coherency.
Figure 6 with 1 supplement
The infection model recapitulates the human neutrophil response to N. meningitidis infection.
(A) Confocal images of E-selectin staining in the Vessel-on-Chip (VoC) for four conditions: without infection nor treatment, after 4 hr of either TNFα treatment or infection with wild-type (WT) or pilD Nm strains. Scale bar: 50 µm. (B–C) Graphs representing the percentage of E-selectin-positive cells and the mean intensity of CD62 in positive cells. For each condition, the mean ± s.d. is represented (Control: 0.32 ± 0.67%–(n=10 vessels), 11.4 ± 1.17 (n=2 cells) — TNFα: 72.0 ± 15.0%–(n=9 vessels), 245 ± 100 (n=244 cells) — WT: 33.1 ± 8.12%–(n=8 vessels), 223 ± 118 (n=126 cells) — pilD: 14.9 ± 8.1%–(n=10 vessels), 144 ± 58.4 (n=44 cells)). (D) Schematic representation of the setup. Purified neutrophils were introduced in the microfluidic chip under flow conditions (0.7–1 µl/min). (E) Bright-field and fluorescence images of neutrophils adhering on a non-treated (control), TNFα-treated, or Nm-infected VoC. Scale bar: 25 µm. (F) Graph representing the number of neutrophils adhering to the endothelium for each condition. Each dot represents a vessel. For each condition, the mean ± s.d. is represented (Control: 0.59 ± 1.32 µm-2 (n=5) — TNFα: 33.0 ± 36.1 µm-2 (n=8) — N. meningitidis: 21.5 ± 28.4 µm-2 (n=6)). (G) Representative images of bacteria phagocytosis by neutrophils in infected VoC (left) and in infected human vessels in the grafted mouse model (right). Scale bar: 25 µm. All statistics have been computed with Wilcoxon tests.
Figure 6—figure supplement 1
The infected Vessel-on-Chip model recapitulates the human neutrophil response to N. meningitidis infection.
(A–C) Brightfield and fluorescence confocal images of a VoC before (-) and after (+) 4 hr of TNFα treatment, or infection of wild-type (WT) or pilD Nm strains (top). Graph of the E-selectin mean fluorescence intensity quantification for each condition (bottom). Each dot represents a vessel. For each condition, the mean ± s.d. is represented (TNFα - Before treatment: 1.85 ± 3.85–4 hr post treatment: 224.0 ± 93.7 (n=17) — WT - Before infection: 1.42 ± 1.02–4 hr p.i.: 105.0 ± 54.4 (n=17) — pilD - Before infection: 1.42 ± 1.02–4 h p.i.: 105.0 ± 54.4 (n=9)) Statistics have been computed with t-tests. (D) Time-lapse of the adhesion cascade of human neutrophils on the human endothelium (VoC). (E) Time-lapse of the adhesion cascade of mouse neutrophils on the mouse endothelium (mouse ear skin).
Author response image 1
F-actin (red) and ezrin (yellow) staining after 3h of infection with N. meningitidis (green) in 2D (top) and 3D (bottom) vessel-on-chip models.
Additional files
Download links
A two-part list of links to download the article, or parts of the article, in various formats.
Downloads (link to download the article as PDF)
Open citations (links to open the citations from this article in various online reference manager services)
Cite this article (links to download the citations from this article in formats compatible with various reference manager tools)
An in vitro human vessel model to study Neisseria meningitidis colonization and vascular damages
eLife 14:RP107813.
https://doi.org/10.7554/eLife.107813.3
