The most striking feature of innate recognition of leptospires is their potent lipoprotein-induced TLR2 activation (112, 114, 122–124). It was first shown that LipL32, the major lipoprotein of leptospires, is a TLR2 agonist in human macrophages and hamster ovary cells (114) and in mouse renal tubular proximal cells (125). We also determined that the recognition of leptospires occurred through dimers of TLR2 and TLR1 (115) specific for triacylated lipoproteins that are present in leptospires (111). More recently, it was shown that leptospires also stimulate pig and bovine fibroblasts through TLR2 (122, 123). The abundant lipoproteins present in the membranes of leptospires easily explain their potent TLR2 activity, at least in vitro. However, the regulation of their expression and posttranslational modifications may lead to immune evasion in vivo (111). Hence, despite being responsible for cellular activation and inflammation in vitro, TLR2 triggering is not essential to mouse defense. Indeed, in contrast with TLR4 knockout (KO) mice, TLR2KO mice do not die from experimental leptospirosis (116, 126). However, activation of TLR2 in addition to TLR4 contributes to host protection in mice by triggering cytokine secretion and leptospire-specific IgG, NO, and IFNγ secretion (116). In addition, we recently showed that the inflammatory response of PMNs infected with L. interrogans was dependent on TLR2 in human neutrophils (55).

Notably, we have also highlighted the potential protective role of TLR2 in the human defense against Leptospira. Indeed, single nucleotide polymorphisms (SNPs) in TLR2 and TLR1, which are known to modulate susceptibility to infectious diseases, have been linked to increased susceptibility to developing severe leptospirosis in Argentina (124). However, the Argentinian cohort of patients was small, and the SNP found in TLR2 has not been associated with increased susceptibility to leptospirosis in a population from the Azores Islands (127).

LPS is a macromolecule composed of three sections: (i) a carbohydrate part responsible for the antigenic properties of LPS, (ii) a central core and KDO region, and (iii) lipid A, consisting of a disaccharide moiety linked to lipid chains that anchor LPS in the outer membrane. Lipid A, also known as “endotoxin,” is recognized by the TLR4/MD2 complex and is responsible for the toxic properties of LPS.

We first showed that the LPS of leptospires is not recognized through TLR4, as expected in human MΦ (114). However, it is recognized by murine TLR4 (115). The structure of leptospiral lipid A has been deciphered by the Raetz group and showed several modifications, including a methylated phosphate group and four amide-linked acyl chains, compared to the structure of classic lipid A from Enterobacteria, and these modifications result in the lack of reactive phosphate groups on each side of the disaccharide (113). These modifications potentially explain the lack of human TLR4 recognition, which is more stringent than its murine counterpart. Using deficient mice, it was shown that TLR4 is a major PRR involved in the control of leptospires (116, 128). The species specificity of TLR4 recognition most likely explains the differences in the severity of leptospirosis, which can be fatal in humans but asymptomatic in mice that are considered chronic carriers of Leptospira (34).

In addition, we recently showed that in mouse macrophages, TLR4 recognition of leptospiral LPS is not complete (117). Only one arm of the signaling pathway is triggered. Indeed, TLR4 has two adaptors: MyD88 at the cell surface and TRIF in the endosome. MyD88 signaling results in cytokine and costimulatory molecule expression, whereas TRIF leads to production of type I interferons and nitric oxide (NO) (117). Leptospiral LPS only stimulates the MyD88 pathway but does not stimulate the TRIF pathway, and as a consequence, whole bacteria induce only minimal antimicrobial NO and IFN-ß production. This mechanism may participate in bacterial evasion and potentially explains why LPS is a virulence factor (129) and why pathogenic leptospires succeed in reaching the mouse kidney despite TLR4 recognition. We showed that the escape mechanism is linked to the presence of a complete O antigen in LPS and tightly bound lipoproteins, which together impair the interaction of LPS with CD14, an important coreceptor required for TLR4 internalization and TRIF signaling (117). Indeed, purification protocols for leptospiral LPS (130) always lead to copurification of TLR2 agonists, suggesting the close structural association of LPS and lipoproteins (114). Moreover, we showed that the protein portion of lipoproteins but not the lipidic portion that signals through TLR2 is important for TRIF escape (117). In line with these data, a recent study showed that one strain of L. interrogans serovar Autumnalis responsible for the self-resolution of infection in a murine model possesses an LPS devoid of TLR2 activity (131).

TLR5 is the receptor of bacterial flagellins, composing the filamentous part of flagella. We previously showed that live leptospires triggered production of equivalent amounts of IL1ß and IL6 cytokines in wild-type and TLR5KO bone marrow–derived MΦ, suggesting that murine TLR5 does not recognize leptospires (132). However, it has also been shown in human whole blood infected or stimulated with inactivated leptospires that a neutralizing TLR5 antibody decreased the cytokine response, suggesting that human TLR5 may recognize leptospires (133). Both observations were confirmed and explained by our recent study showing that live leptospires escape TLR5 recognition in both human and mouse macrophages, but leptospires that are either degraded by heating or killed by antimicrobial peptides such as the cathelicidins LL37 or bMap28 are sensed by human and bovine TLR5, although they are not detected or only barely detected by mouse TLR5. These results confirmed the hypothesis that the periplasmic localization of the endoflagella participates in immune evasion. From a therapeutic perspective, since Enterobacteria flagellin is a potent adjuvant (134), one may also hypothesize that escape from TLR5 could potentially limit the adaptive immune response to Leptospira. Moreover, we also highlighted that the structure of leptospiral FlaB flagellins, which are agonists of TLR5, is devoid of the variable portion that confers the antigenicity of flagellins, suggesting that leptospiral flagellins would not be efficiently recognized by antibodies; this adds a second layer of immune evasion.

NOD1 and NOD2 are cytosolic PRRs that recognize bacterial fragments of peptidoglycan called muropeptides, which are constantly released by the bacteria upon remodeling and synthesis of the cell wall (9). NOD1 and NOD2 are important for controlling invasive and extracellular bacteria (9). Activation of NOD receptors is also important for immune functions and the onset of adaptive immune responses (112). Indeed, in mice, NOD1 activation by the microbiota has been shown to be crucial for PMN killing of bacteria and fungi (135). In addition, we showed that NOD1 activation plays an important role in murine renal defense using a model of retro-urethral infection with uropathogenic Escherichia coli (136).

We showed that leptospires escape NOD1 and NOD2 recognition. We demonstrated that the lipoprotein LipL21 is tightly bound to peptidoglycan and impairs the release of muropeptides, therefore blocking NOD1/2 sensing (121). Moreover, we found a species-specific mechanism of NOD1 escape. Human and mouse NOD1 do not recognize exactly the same agonists (9). We showed that the peptidoglycan of L. interrogans is devoid of the murine NOD1 agonist but possesses fair amounts of the human NOD1 agonist, suggesting that bacteria in which LipL21 would be degraded or missing could also escape NOD1 sensing in mice but would still signal in human cells. This hypothesis has been confirmed using humanized NOD1 mice infected with the L. interrogans Manilae L495 and lipL21 mutants (121).

The other role of LipL21, its potent inhibition of MPO activity in PMNs (60), is probably not linked to PG binding and NOD1 escape, since experiments with recombinant LipL21 led to similar observations as experiments with membrane fractions of whole bacteria (60).

Therefore, we hypothesize that LipL21 blocks muropeptide release (121) and, along with LipL45, inhibits MPO (60), most likely contributing to neutrophil escape and immune evasion. Because the leptospiral genome encodes more than 140 lipoproteins, most of them with unknown functions (111), we may speculate that other lipoproteins participate in immune evasion.

The inflammasome is a platform for cytosolic proteins composed of NOD-like protein (NLRP), one or several adaptors such as the protein ASC, and pro-caspase 1. Inflammasome activation results in the secretion of pro-inflammatory IL1β. This cytokine is central to inflammation and is tightly regulated. In mice, two signals are required; the first signal leads to NF-κB activation and mRNA expression of pro-IL1ß and the components of the inflammasome, and the second signal activates the inflammasome through oligomerization of the NLRP proteins and recruitment of the ASC adaptor and pro-caspase 1, which itself is cleaved and activated to convert pro-IL1ß into mature IL1ß (9).

We have shown in murine bone marrow-derived MΦ that leptospires trigger the NLRP3 inflammasome through LPS and lipoprotein activation of TLR2 and TLR4 and downregulation of the potassium pump by glycolipoprotein (GLP) (119). We also showed that leptospiral activation of the inflammasome was mostly limited to NLRP3 and excluded other potential inflammasomes, such as NLRC4/NAIP5, which recognize flagellin. NLRP3-dependent IL1β secretion was confirmed in a recent study showing that doxycyclin reduced IL1β secretion in J774 murine MΦ cells infected with leptospires (137). Subsequently, it was shown that leptospires also trigger the NLRP3 inflammasome in the THP1 human MΦ cell line (120). The mechanism of activation was not investigated but revealed the contribution of ROS and cathepsin B, which was not found in mice.

One key point to consider is the species specificity of the innate immune responses to leptospiral MAMPs. Several studies presented in this review highlighted differences in the recognition of leptospires between different hosts that could contribute to species-specific sensitivity to the disease (Figure 4). For example, the intracellular fate of leptospires in phagocytes (20) as well as TLR4, TLR5 and NOD1 recognition (115, 117, 118, 121) differ between humans and mice. Recent studies on neutrophil and platelet activation by leptospires also suggest differences between humans and rats or rabbits (54–56, 64). In addition, we also highlighted a difference in TLR5 sensing in mice, in which TLR5 does not sense leptospiral flagellins, and cows, in which TLR5 can sense them, as in humans (118). Recognition of PRRs of leptospires by TLRs involved in nucleic sensing, including TLR3, TLR7, TLR8, TLR9, and those recognizing microbial RNAs and DNA, as well as other cytosolic sensors should also be considered.

Several recent studies from the Cao group used PRR agonists at the time of infection or 1 day prior to the infection, which showed some protective effects in hamsters, a sensitive animal model of leptospirosis (34).

First, it was shown that early expression of TLR2 (but not TLR4) was observed upon leptospiral infection in resistant BALB/c mice, whereas delayed expression was observed in sensitive hamsters (138). This confirmed a previous study from Matsui et al., suggesting that the early triggering of inflammatory mediators in mice infected with leptospires was protective, whereas the delayed response in hamsters could participate in a cytokine storm (140). Interestingly, anti-inflammatory IL-10 cytokine secretion was shown to be dependent on TLR2 in mice (140). The coinjection of hamsters with L. interrogans serovar Autumnalis and Pam3cysSK4, a synthetic agonist of TLR2, alleviated acute leptospirosis and improved the survival of hamsters, which showed reduced leptospiral loads and histopathological lesions in organs 3 weeks pi (138).

More recently, the same group investigated the effect of crude E. coli LPS administered in hamsters after infection with leptospires and showed an improved outcome in hamsters, which was correlated with increased inflammation levels (126). However, the use of nonrepurified crude E. coli LPS, which is known to activate both TLR2 and TLR4 (126, 141), made the interpretation of the relative contributions of TLR4 and TLR2 difficult. Nonetheless, an interesting observation is that upon leptospiral infection, cotreatment with E. coli LPS dramatically enhanced both NO secretion and MPO activity. However, in vivo pharmacological inhibition of NO did not impact the outcome due to LPS (119), suggesting that the enhanced or restored MPO activity (or activity of other mediators) could be responsible for protection. Interestingly, coinjections starting 1 day prior to the time of leptospiral infection with ß-glucan, a fungal cell wall component agonist of the Dectin-1 receptor known to synergize with TLR2, also increased early inflammation and improved survival (139).

By attempting to use Lactobacillus plantarum, a generally recognized as safe (GRAS) organism, as a platform to express leptospiral proteins to immunize sensitive C3H/HeJ mice through the oral route, we observed unanticipated protective effects. Indeed, after a 6-week regimen of intermittent oral gavage, it was shown that treatment with the parental L. plantarum strain protected mice from acute leptospirosis in the sensitive C3H/HeJ TLR4-deficient model (142). The symptoms of leptospirosis were alleviated, and 15 days postleptospiral challenge, the effect of L. plantarum was associated with the recruitment of macrophage-like cells in lymphoid organs and kidneys (142). L. plantarum did not prevent renal colonization, but the treatment reduced inflammation and renal fibrosis. Because these features were reminiscent of the newly introduced concept of “innate immune memory”, also known as “trained immunity,” during which a first infection or MAMP stimulation triggers metabolic and epigenetic reprogramming of MΦ and NK cells to cause them to more robustly respond upon a second challenge, we hypothesized that L. plantarum triggered such a mechanism. We tested the effect of two intraperitoneal injections of CL429, a dual TLR2-NOD2 agonist that was shown by others to recapitulate the protective and anti-inflammatory effects of oral L. plantarum in mice in a model of viral pulmonary infection (143). We demonstrated that CL429 pretreatment indeed induced an innate memory effect independent of B and T cells in MΦ from the peritoneal cavity but also in distant sites, such as the bone marrow and NK cells from spleen (17). CL429 pretreatment enhances the MΦ secretion of cytokines and NO and IFNγ production by NK cells. Notably, the CL429 effect lasted for at least 3 months posttreatment and was effective against the 3 L. interrogans serovars tested (17) (Figure 5). In addition, CL429 also triggered a trained immunity effect in human MΦ derived from monocytes of healthy blood donors (17). A recent study showed that ß-glucan, the prototypic MAMP used to induce trained immunity, was consistently effective in preventing severe leptospirosis in hamsters when administered 5 days prior to infection with L. interrogans serovar Lai strain 56601 (139) (Figure 5). ß-glucan protected 37.5% of hamsters from death and alleviated lesions in the liver, lungs, and kidneys (139).