Escape of TLR5 Recognition by Leptospira spp.: A Rationale for Atypical Endoflagella

Abstract

Leptospira (L.) interrogans are invasive bacteria responsible for leptospirosis, a worldwide zoonosis. They possess two periplasmic endoflagellae that allow their motility. L. interrogans are stealth pathogens that escape the innate immune recognition of the NOD-like receptors NOD1/2, and the human Toll-like receptor (TLR)4, which senses peptidoglycan and lipopolysaccharide (LPS), respectively. TLR5 is another receptor of bacterial cell wall components, recognizing flagellin subunits. To study the contribution of TLR5 in the host defense against leptospires, we infected WT and TLR5 deficient mice with pathogenic L. interrogans and tracked the infection by in vivo live imaging of bioluminescent bacteria or by qPCR. We did not identify any protective or inflammatory role of murine TLR5 for controlling pathogenic Leptospira. Likewise, subsequent in vitro experiments showed that infections with different live strains of L. interrogans and L. biflexa did not trigger TLR5 signaling. However, unexpectedly, heat-killed bacteria stimulated human and bovine TLR5, but did not, or barely induced stimulation via murine TLR5. Abolition of TLR5 recognition required extensive boiling time of the bacteria or proteinase K treatment, showing an unusual high stability of the leptospiral flagellins. Interestingly, after using antimicrobial peptides to destabilize live leptospires, we detected TLR5 activity, suggesting that TLR5 could participate in the fight against leptospires in humans or cattle. Using different Leptospira strains with mutations in the flagellin proteins, we further showed that neither FlaA nor Fcp participated in the recognition by TLR5, suggesting a role for the FlaB. FlaB have structural homology to Salmonella FliC, and possess conserved residues important for TLR5 activation, as shown by in silico analyses. Accordingly, we found that leptospires regulate the expression of FlaB mRNA according to the growth phase in vitro, and that infection with L. interrogans in hamsters and in mice downregulated the expression of the FlaB, but not the FlaA subunits. Altogether, in contrast to different bacteria that modify their flagellin sequences to escape TLR5 recognition, our study suggests that the peculiar central localization and stability of the FlaB monomers in the periplasmic endoflagellae, associated with the downregulation of FlaB subunits in hosts, constitute an efficient strategy of leptospires to escape the TLR5 recognition and the induced immune response.

Keywords: Flagelin genes; Leptospira; TLR5; innate immunity; mouse model; toll-like receptor.

FIGURE 1

No difference in leptospiral loads and inflammatory mediators between WT and TLR5ko mice during acute phase of leptospirosis. (A,B) Bacterial loads determined by qPCR of leptospiral DNA at several days post intraperitoneal infection of 7-week old female C57BL/6J (WT) mice and TLR5ko mice with 10⁷ L. interrogans Manilae L495 derivative strain MFLum1; (A) in blood (red) and urine (yellow) in WT (n = 4, hatched bars) and TLR5ko mice (n = 4, empty bars) and (B) in liver (Li), spleen (Sp), lungs (Lu) and kidneys (Ki) from WT (n = 4, black bars) and TLR5ko (n = 4, blue bars) at day 3 post-infection (p.i). Data are expressed as mean (±SD) of all mice. Statistically significant differences between genotypes (Student t-test) are indicated. (C) Bacterial loads in kidneys determined by qPCR 7 days p.i of WT mice (n = 5, black dots) and TLR5ko mice (n = 7, blue dots). (D) Inflammation measured in kidney by mRNA expression of cytokines (RANTES, IL10, IFNγ) measured by RT-qPCR at 7 days p.i in WT mice (n = 5, black dots) and TLR5ko mice (n = 7, blue dots). Individual mice are represented as dots and lines correspond to mean (±SD) of all mice. No statistical difference was found between WT and TLR5ko mice. *p < 0.05.

FIGURE 2

No difference in kidney colonization in WT and TLR5ko mice 15 days post-infection. (A) Bacterial loads in urine quantified by qPCR (left panel), and quantification and images of live imaging (IVIS) (right panel) 15 days post-infection (p.i) of 7-week old female WT mice (n = 7, black) and TLR5ko mice (n = 8, blue) with 10⁷ L. interrogans Manilae bioluminescent derivative strain MFLum1. Imaging was performed in dorsal position, 24 h post shaving, on anesthetized mice and after luciferin administration. The background level of light after luciferin administration was measured on a control TLR5ko mouse injected with PBS at the time of infection (dotted line). The average radiance in individual mice gated on the whole body is shown and represented as dots; lines correspond to the mean (±SD) of all mice. No statistical difference was found between WT and TLR5ko mice. (B) Bacterial loads in kidneys determined by qPCR of leptospiral DNA 14 days p.i of WT mice (n = 6, back dots) and TLR5ko (n = 6, blue dots) mice with 10⁷L. interrogans Copenhageni Fiocruz L1-130. Individual mice are shown and represented as dots; lines correspond to mean (±SD) of all mice. No statistical difference was found between WT and TLR5ko mice. (C) Histological sections and immunolabeling of the kidneys of naive TLR5ko, infected WT and TLR5ko mice 15 days p.i with 10⁷ L. interrogans Manilae strain MFLum1. (a–c) Kidney, Hematoxylin-Eosin stain, Original magnification ×2, Scale bar: 500 μm. Cortex (Co), Medulla (Me), Papilla (Pa), Capsule (Ca). (d–f) Kidney cortex, Hematoxylin-Eosin stain, Original magnification ×10, Scale bar: 100 μm. The asterisks indicate the focal inflammatory infiltrates. (g–i) Anti-LipL21 labeling of leptospires in renal tubules, Original magnification ×10, Scale bar: 100 μm. (j,k) Double labeling LipL21/Periodic Acid-Schiff (PAS) to stain the PAS positive brush borders present in proximal tubules only. Original magnification ×40, Scale bar: 25 μm.

FIGURE 3

Heat-killed, but not live leptospires, signal through hTLR5. (A,C) KC production measured by ELISA in the supernatants of BMMs from WT (black bars) and TLR5ko (blue bars) mice 24 h post-infection with MOI 50 of either (A) live or (C) heat-killed (30 min, 100°C) different serovars of virulent L. interrogans (Manilae strain L495, Copenhageni strain Fiocruz L1-130, Icterohaemorrhagiae strain Verdun). LPS from E. coli (100 ng/mL) and unpurified Fla from Salmonella typhimurium (500 ng/mL) were used as controls. Data are expressed as mean (±SD) of technical replicates (n = 5) on pooled BMMs preparations from mice (n = 3) and are representative of at least three independent experiments. Statistically significant differences between genotypes (Student t-test) are indicated. (B,D) NF-κB reporter assay in HEK-Blue-Knock Down (KD)-TLR5 cells transfected with the mouse TLR5 (light blue bars), human TLR5 (blue bars), or empty plasmid (empty bars) and stimulated for 24 h with MOI 200 of either (B) live or (D) heat-killed (30 min, 100°C) different serovars of virulent L. interrogans. Unpurified Fla from Salmonella typhimurium (100 ng/mL) was used as control. Data are expressed as the mean (±SD) of technical replicates (n = 3) and are representative of at least three independent experiments. Statistically significant differences (Student t-test) are indicated. *p < 0.05; ***p < 0.001.

FIGURE 4

A very stable protein from leptospires signal through TLR5. (A) NF-κB reporter assay in HEK-Blue-KD-TLR5 cells transfected with the human TLR5 (blue bars), or empty plasmid (empty bars) and stimulated with MOI 100 of either live or heat-killed (30 min, 100°C) L. interrogans Copenhageni strain Fiocruz L1-130 treated or not with Proteinase K (protK) followed or not by heat inactivation at 99°C for 30 min (inact or non-inact). Unpurified Fla from Salmonella typhimurium (100 ng/mL) was used as control. Data are expressed as the mean (±SD) of technical replicates (n = 3) and are representative of at least three independent experiments. Statistically significant differences (Student t-test) are indicated. (B) Chronogram of proteinase K experiments. (C) Picture of NF-κB reporter assay in HEK-Blue-KD-TLR5 cells transfected with the human TLR5 or empty plasmid and stimulated with MOI 100 of live L. interrogans Copenhageni strain Fiocruz L1-130 incubated at various temperatures during 30 min, 3, or 8 h. Unpurified Fla from Salmonella typhimurium (500 ng/mL) was used as control. Picture show technical duplicate for each condition and is representative of at least three independent experiments. (D) NF-κB reporter assay in HEK-Blue-KD-TLR5 cells transfected with the human TLR5 (blue bars) or empty plasmid (empty bars) and stimulated with MOI 100 of live L. interrogans Copenhageni strain Fiocruz L1-130 or Manilae L495 incubated at various temperatures during 3 h. Unpurified Fla from Salmonella typhimurium (500 ng/mL) was used as control. Data are expressed as the mean of technical replicates (n = 2, shown as dots) and are representative of at least three independent experiments for Fiocruz L1-130 and two independent experiments for L495. ***p < 0.001.

FIGURE 5

Human and bovine antimicrobial peptides unmask the leptospiral ability to signal through human and bovine TLR5 receptors. (A) Alamar blue viability assay of leptospires (Manilae L495 or Copenhageni Fiocruz L1-130) incubated with increasing concentration (0–250 μg/mL) of antimicrobial peptides LL-37 or Bmap28 for 2 h. Heat-killed (30 min, 100°C) leptospires are used as controls for loss of viability. Picture show technical triplicate for each condition and is representative of two independent experiments. (B) NF-κB reporter assay in HEK-Blue-KD-TLR5 cells transfected with the human TLR5 (blue bars), or empty plasmid (empty bars) and stimulated with MOI 100 of L. interrogans Manilae strain L495 or Copenhageni strain Fiocruz L1-130 treated with human peptide LL-37 or bovine peptide Bmap28 at various concentration (0–250 μg/mL) for 2 h before stimulation. (C) NF-κB reporter assay in HEK-Blue-KD-TLR5 cells transfected with the human TLR5 (blue bars), bovine TLR5 (dark blue bars), mouse TLR5 (light blue bars) or empty plasmid (empty bars) and stimulated with MOI 100 of either live or heat-killed (30 min, 100°C) L. interrogans Manilae strain L495 or Copenhageni strain Fiocruz L1-130. (D) NF-κB reporter assay in HEK-Blue-KD-TLR5 cells transfected with the bovine TLR5 (dark blue bars), mouse TLR5 (light blue bars) or empty plasmid (empty bars) and stimulated with MOI 100 of L. interrogans Copenhageni Fiocruz L1-130 treated with human peptide LL-37 or bovine peptide Bmap28 at various concentration (0–250 μg/mL) for 2 h before stimulation. (B–D) Unpurified Fla from Salmonella typhimurium (500 ng/mL) was used as control. Data are expressed as the mean (±SD) of technical replicates (n = 3), and are representative of at least three independent experiments for panels (A) and (C). Statistically significant differences (Student t-test) are indicated. *p < 0.05; **p < 0.01; ***p < 0.001.

FIGURE 6

Comparison of leptospiral Flagellins and FliC structures in relation with TLR5. (A) Amino acid sequence homology average percentage between Salmonella typhimurium FliC (P06179) and Leptospira interrogans strain Fiocruz FlaB (LIC11890, LIC11889, LIC11532 and LIC11531) and FlaAs (LIC10788 and LIC10787) and primary structures of the flagellin proteins with TLR5 binding consensus. (B) In silico (Phyre2 and Chimera softwares) prediction of Salmonella typhimurium FliC (P06179) structure with the four described domains and with positions of the TLR5 binding consensus: 1 (red), 2 (yellow) and 3 (light blue) and stabilization region (light green) highlighted. (C) In silico (Phyre2 and Chimera softwares) prediction of Leptospira interrogans strain Fiocruz FlaB4 (LICI1531) with the positions of the TLRS binding consensus and stabilization region highlighted, FlaA1 (LIC10788), FlaA2 (LICI0787). (D) Clustal (MEGA software) alignment of the amino acid sequences for the TLR5 binding consensus regions of: Salmonella enterica FliC (GeneBank QDQ31983.1), L. bifleva (strain Patoc) FlaB1 (LEPBla2133), FlaB2 (LEPBIa2132), FlaB3 (LEPBla1872) and FlaB4 (LEPBla1589), L. interrogans (strain Fiocruz L1-130) FlaB1 (LIC18890), FlaB2 (LICIT889), FlaB3 (LICI1532) and FlaB4 (LIC11531), L. interrogans (strain L495) FlaB1 (LMANv2 260016), FlaB2 (LMANv2 260015). FlaB3 (LMANv2 590024) and FlaB4 (LMANv2 590023), and L. interrogans (strain Verdun) FlaB1 (AKWP_v1_110429), FlaB2 (AKWP_v1_110428) and FlaB3 (AKWP_v1_110068) and FlaB4 (AKWP_v1_110067).

FIGURE 7

FlaB subunits, not FlaAs nor Fcps, contribute to the signaling. (A) NF-κB reporter assay in HEK-Blue-KD-TLR5 cells transfected with the human TLR5 (blue bars), or empty plasmid (empty bars) and stimulated with MOI 100 of either live or heat-killed (30 min, 100°C) L. interrogans Copenhageni Fiocruz LV2756 WT or ΔFcpA, Patoc Patoc I WT or ΔFcpA. Unpurified Fla from Salmonella typhimurium (500 ng/mL) was used as control. Data are expressed as the mean (±SD) of technical replicates (n = 3) and are representative of at least three independent experiments. (B) NF-κB reporter assay in HEK-Blue-KD-TLR5 cells transfected with the human TLR5 (blue bars), or empty plasmid (empty bars) and stimulated with MOI 100 of either live or heat-killed (30 min, 100°C) L. interrogans Manilae L495 WT or ΔFlaA2. Unpurified Fla from Salmonella typhimurium (500 ng/mL) was used as control. Data are expressed as the mean of technical duplicates (n = 2, shown as dots). (C) NF-κB reporter assay in HEK-Blue-KD-TLR5 cells transfected with the human TLR5 (blue bars), or empty plasmid (empty bars) and stimulated with MOI 100 of either live or heat-killed (30 min, 100°C) L. interrogans Manilae L495 WT or ΔFlaB1 and Patoc Patoc I WT or ΔFlaB4. Data are expressed as the mean (±SD) of technical replicates (n = 3) and are representative of at least three independent experiments. Statistically significant differences (Student t-test) are indicated. **p < 0.01; ***p < 0.001.

FIGURE 8

FlaBs mRNA are upregulated in stationary phase and downregulated in vivo. (A) In vitro FlaBs mRNA expression in L. interrogans Copenhageni Fiocruz L1-130, Icterohaemorrhagiae Verdun and Manilae L495 at the exponential (E) and stationary (S) phase. Data of RT-qPCR are expressed as the relative mRNA quantities normalized to the expression of the lipl41 mRNA. Technical replicates are represented as dots and lines correspond to mean (±SD) of replicates (3 < n < 9). Statistically significant differences (Student t-test) are indicated. (B) In vivo FlaAs and (C) FlaBs mRNA expression in blood of infected mice (n = 5, light blue) and hamsters (n-5, dark blue), 24 h post intraperitoneal infection with 2 × 10⁸ virulent L. interrogans Icterohaemorrhagiae strain Verdun, compared with mRNA expression in culture in EMJH at 30°C. Data of RT-gPCR are expressed as the ratio of mRNA quantities relatives to the EMJH control. Individual animals are represented as dots and lines correspond to mean (±SD) of all animals. Statistically significant differences (Student t-test) are indicated with corresponding p values: * for p < 0.05; ** for p < 0.01 and *** for p < 0.001.