Type IV pilus retraction enables sustained bacteremia and plays a key role in the outcome of meningococcal sepsis in a humanized mouse model

Abstract

Neisseria meningitidis (the meningococcus) remains a major cause of bacterial meningitis and fatal sepsis. This commensal bacterium of the human nasopharynx can cause invasive diseases when it leaves its niche and reaches the bloodstream. Blood-borne meningococci have the ability to adhere to human endothelial cells and rapidly colonize microvessels. This crucial step enables dissemination into tissues and promotes deregulated inflammation and coagulation, leading to extensive necrotic purpura in the most severe cases. Adhesion to blood vessels relies on type IV pili (TFP). These long filamentous structures are highly dynamic as they can rapidly elongate and retract by the antagonistic action of two ATPases, PilF and PilT. However, the consequences of TFP dynamics on the pathophysiology and the outcome of meningococcal sepsis in vivo have been poorly studied. Here, we show that human graft microvessels are replicative niches for meningococci, that seed the bloodstream and promote sustained bacteremia and lethality in a humanized mouse model. Intriguingly, although pilus-retraction deficient N. meningitidis strain (ΔpilT) efficiently colonizes human graft tissue, this mutant did not promote sustained bacteremia nor induce mouse lethality. This effect was not due to a decreased inflammatory response, nor defects in bacterial clearance by the innate immune system. Rather, TFP-retraction was necessary to promote the release of TFP-dependent contacts between bacteria and, in turn, the detachment from colonized microvessels. The resulting sustained bacteremia was directly correlated with lethality. Altogether, these results demonstrate that pilus retraction plays a key role in the occurrence and outcome of meningococcal sepsis by supporting sustained bacteremia. These findings open new perspectives on the role of circulating bacteria in the pathological alterations leading to lethal sepsis.

Fig 1. Human graft is a replicative niche for Neisseria meningitidis allowing sustained bacteremia.

(A) Grafted mice were infected with a rising range of inocula: 1x10¹, 5x10¹, 5x10², and 5x10³ CFU of WT N. meningitidis. Bacteremia and graft bacterial load were assessed at the time of sacrifice 18 h PI by quantitative culture. Bacterial counts are expressed in CFU/ml for blood and in CFU/g for graft. Two independent experiments, n = 5 or 6 mice per group, grafted with skin obtained from two different donors. Bars represent mean ± SEM. (B) Grafted mice were infected intravenously with 5x10⁶ CFU of WT N. meningitidis and its isogenic piliated non-adhesive mutant ΔpilC1. Bacteremia was measured at 1, 18, 48 and 72 h by culturing serial dilutions of blood samples on agar plates. Two independent experiments, n = 6 or 8 mice per group. Bars represent mean ± SEM, **** p < 0.0001, one-way ANOVA followed by multiple comparison test. (C) Kaplan-Meier plot showing the survival of infected grafted mice shown in panel B. Mice survival was assessed each day during 72 h. Two independent experiments, ** p < 0.01, two-sided log-rank Mantel-Cox analysis. (D) A competition assay between WT N. meningitidis and its isogenic piliated non-adhesive mutant ΔpilC1 was performed on both grafted (left panel) and non-grafted (right panel) mice. Mice were infected IV with an inoculum of 1x10⁷ CFU total (a mixture of 5x10⁶ CFU of WT and 5x10⁶ CFU of ΔpilC1 mutant). Competitive index, defined as the mutant/WT ratio within the output sample, divided by the corresponding ratio in the inoculum, was measured in the blood of mice at 1 h and 18 h PI by tail vein blood puncture. Two independent experiments, n = 8 mice per group. Bars represent mean ± SEM, NS p > 0.05; ** p < 0.01; **** p < 0.0001, one-way ANOVA followed by multiple comparison test.

Fig 2. Pilus-retraction is required to establish sustained bacteremia and lethality.

(A) Grafted mice were infected IV with 5x10⁶ CFU of N. meningitidis, ΔpilT mutant and its complemented Cp-pilT strain. Bacteremia of mice was measured at 1, 18, 48 and 72 h PI by culturing serial dilutions of blood samples on agar plates. Two independent experiments, n = 8 or 9 mice per group. Bars represent mean ± SEM, * p < 0.05, one-way ANOVA followed by multiple comparison test. (B) Kaplan-Meier plot showing the survival of infected grafted mice shown in panel A. Mice survival was assessed each day during 72 h. Two independent experiments, ** p < 0.01, two-sided log-rank Mantel-Cox analysis. (C) Grafted mice were infected IV with 5x10⁶ CFU of WT N. meningitidis, isogenic ΔpilT mutant, or complemented strain Cp-pilT. Graft bacterial load at 18 hours PI was measured by quantitative culture on agar plates. Two independent experiments for WT and ΔpilT strains and one experiment for complemented Cp-pilT strain, n = 3 or 6 mice per group. Bars represent mean ± SEM, NS p > 0.05, one-way ANOVA followed by multiple comparison test.

Fig 3

N. meningitidis WT and ΔpilT induce equivalent inflammatory responses (A, B) Serum levels of proinflammatory mouse and human cytokines were measured in grafted mice infected IV with 5x10⁶ CFU of N. meningitidis WT strain or isogenic ΔpilT mutant at 4 h (A) and 18 h (B) post-infection using multiplex assays. Two independent experiments, n = 4, 5 or 6 mice per group. Bars represent mean ± SEM, NS p > 0.05, unpaired t-test.

Fig 4. N. meningitidis WT and ΔpilT are similarly phagocytized.

(A, B) N. meningitidis phagocytosis at 4 h and 18 h PI in the blood or spleen of grafted (A) and non-grafted (B) SCID mice. Neutrophils were identified as CD11b⁺Ly6G^hi. Red pulp macrophages (RPM) were identified as CD11b^neg F4/80^hi and proinflammatory monocytes were identified as CD11b⁺ Ly6G⁻ Ly6C^hi F4/80⁺. Two independent experiments for grafted mice with n = 6 mice in each group (B) and one experiment for non-grafted mice with n = 3 mice in each group (A). Bars represent mean ± SEM, NS p > 0.05; * p < 0.05, one-way ANOVA followed by multiple comparison test.

Fig 5. Monocytes and neutrophils actively control the bacteremia induced by WT or ΔpilT strains.

(A) Grafted and (B) non-grafted SCID mice treated with cyclophosphamide (cyclo) and infected with 5x10³ CFU of WT N. meningitidis or ΔpilT mutant. Bacteremia was assessed at 18 h post-infection by blood puncture. Mice were sacrificed at 72 h post-infection and graft bacterial load was measured by quantitative culture. Two independent experiments, with n = 6 mice in each group. Bars represent mean ± SEM. NS p > 0.05; * p < 0.05, one-way ANOVA followed by multiple comparison test.

Fig 6. PilT-dependent pilus retraction is critical for bacterial release from microcolonies adherent to endothelial cells.

(A) Aggregates of WT N. meningitidis, its isogenic ΔpilT mutant, Cp-pilT complemented strain and ΔpilX mutant producing GFP under an IPTG-inducible promoter and grown in liquid medium were imaged using the IncuCyte S3 platform at 37°C with 5% CO₂. Four image sets from distinct regions per well were taken every 15 min using a ×20 dry objective and each condition was run in triplicate. Bacterial aggregates formation and dispersion was measured over a period of 20 h under static condition. (B) Number of bacteria detaching from microcolonies adhering onto human dermal microvascular cells (HDMEC) under flow condition. Cells were infected for 30 min under static condition with WT N. meningitidis, ΔpilT mutant, Cp-pilT complemented strain, and ΔpilC1 mutant as non-adhesive control. Cells were then washed to remove non-adherent bacteria and placed under shear stress of 0.15 dyn/cm². Bacterial count in the effluent was evaluated each hour by plating serial dilutions on agar plates. Three independent experiments. Bars represent mean ± SEM. * p < 0.05, one-way ANOVA followed by multiple comparison test.

Fig 7. Maintenance of a sustained high bacteremia correlates with lethality.

(A, B) Non-grafted SCID mice were infected IV with N. meningitidis WT strain (A) or ΔpilT mutant (B). A first infection was performed with an inoculum of 3x10⁶ CFU. A second infection with the same inoculum was performed in the 2 injections group 8 h after the first injection to mimic a prolonged bacteremia (2 injections), as control mice received one injection (1 injection). Mice survival was assessed each day during 72 h. Two independent experiments with n = 6 mice in each group. Kaplan-Meier plots showing the survival of infected mice (left panel, **p < 0.01; *** p < 0.001, two-sided log-rank Mantel-Cox analysis). Bacteremia was measured at 1, 8, 18, 48 and 72 h by culturing serial dilutions of blood samples on agar plates (right panel, bars represent mean ± SEM, **** p < 0.0001, one-way ANOVA followed by multiple comparison test).