Leptospira spp (Leptospira) are not considered gram-negative bacteria but they produce lipopolysaccharide (LPS), a potent inflammatory cell wall component that has been shown to be less inflammatory in Leptospira. Leptospiral-LPS stimulates mouse and human toll like receptor 2 (TLR2) through abundant associated lipoproteins (1). Differences of recognition of leptospiral-LPS exist between mouse and human TLR4, in that the anchor of LPS in the outer membrane of Leptospira (LipidA) activates murine, but not human TLR4 (1, 2). This could be due to the lack of phosphate groups in Leptospira’s LPS lipid A (3). Escape from human TLR4 may allow for lack of immune recognition, which may favor the leptospiral infection.

Researchers working on the function of the host TLR4 in leptospirosis used immunocompetent C57BL/6J and its respective congenic TLR4 knockouts. They found that wild-type (WT) mice expressing competent TLR4 receptors in their immune cells are resistant to lethal infection with L. interrogans whereas TLR4 knockouts are susceptible, have larger numbers of Leptospira in tissues and succumb to infection (4). Furthermore, we and others (5–8) developed mouse models of lethal and sublethal leptospirosis using C3H-HeJ mice that have a point mutation in the cytoplasmic domain of the tlr4 gene (9). These data strongly suggest that the competence of the TLR4 receptor affects leptospirosis outcomes.

Our hypothesis was that if human TLR4 does not sense leptospiral-LPS (1), then transgenic mice in which murine TLR4 was replaced with human TLR4 (along with its co-receptor MD-2) should be more susceptible to Leptospira dissemination as compared to congenic wild-type mice expressing murine TLR4, which could result in a more robust inflammatory response and pathology in the kidney. In this study we compared susceptibility to leptospirosis in C57BL/6J wild-type (WT) and TLR4 humanized transgenic (huTLR4) mice and evaluated signs of disease progression, as well as pathology and inflammation as we previously described (8, 10). As a control we used a congenic mutant strain in which the tlr4 coding sequences are deleted (muTLR4^Lps-del), as we expected it to develop leptospirosis as shown for other C57BL/6J TLR4^KO mice (4).

To find out if replacing the murine TLR4 with human TLR4 affects the kinetics of leptospirosis progression, we compared infection in wild-type C57BL/6J (WT) and humanized TLR4/MD2 transgenic mice (huTLR4) and added a congenic murine mutant lacking the TLR4 receptor as control (muTLR4^Lps-del). We inoculated mice intraperitoneally with a high dose (~10⁸) of L. interrogans Copenhageni FioCruz and analyzed weight loss as a clinical outcome of disease as well as Leptospira dissemination and colonization of the kidney ( Figures 1 and 2 ). In contrast to PBS controls, both WT and huTLR4 slowly lost small but significant amounts of weight over two weeks of infection ( Figures 1A, B ). The muTLR4^Lps-del lost an abrupt and considerably higher amount of weight after day 6 when colonization of the kidney takes place ( Figure 1C ). Differences in weight loss between the three infected groups were significant with huTLR4 being the group that lost less weight ( Figure 1D ).

When we quantified the burden of Leptospira in fluids and tissues by qPCR we found low numbers of Leptospira in blood in WT and huTLR4 mice (mean ~230 and ~300, respectively) and, as expected, a significantly higher number of Leptospira in muTLR4 ^Lps-del (mean ~1200) ( Figure 2A ). Furthermore, much higher numbers of Leptospira (~2-4 Log) were excreted in urine of all infected mice, ~ 41,000 in WT, ~74,000 in huTLR4 and ~305,000 in muTLR4^Lps-del ( Figure 2B ). At termination at 15dpi, tissue PCR showed the same trend with a significantly higher presence of Leptospira in the kidney of muTLR4^Lps-del ( Figure 2C ), and culture of the tissue showed that live Leptospira colonized the kidney of all groups of infected mice ( Figure 2D ). These results suggest that Leptospira infection leads to persistent sublethal leptospirosis, renal colonization and stable shedding of Leptospira in urine of WT and huTLR4 mice.

To find out if replacing murine TLR4 with human TLR4 affects inflammation of the kidney after infection with L. interrogans, we analyzed the differences in histopathology and inflammatory markers in tissue collected from infected and uninfected mice two weeks post infection ( Figures 3 and 4 ). Examination of H&E sections of kidney showed that of all three groups of infected mice, huTLR4 had considerably less mononuclear lymphocyte infiltrates and less tubular damage than WT and muTLR4^Lps-del ( Figure 3A ). We purified mRNA from infected and uninfected kidney. Quantification of expression of iNOS and fibroblast activation marker collagen 1 A1 (ColA1) showed an overall lower detection of iNOS in huTLR4 as well as lower ColA1 in WT and huTLR4 than in muTLR4^Lps-del. Differences between infected groups were statistically significant for ColA1 ( Figure 3B ).

Given that these results suggest that inflammation may not be triggered in kidney of huTLR4 mice after Leptospira infection, we quantified a number of chemokines and cytokines in this tissue ( Figure 4 ). Between infected and uninfected mice, the expression of CxCL1/KC, CxCL2/MIP-2, CCL5/RANTES, TNF-α, IFN-γ; and IL-10 were generally significantly increased in the infected kidneys of WT, huTLR4 and muTLR4^Lps-del control, with the notable exception of CxCL1/KC chemokine and IL-10 in huTLR4 ( Figure 4 ). Between the three groups of infected mice, differences between CxCL1/KC, CxCL2/MIP2, TNF-α and IL-10 were significant. These results suggest that inflammation in kidneys of huTLR4 mice is reduced compared to WT and muTLR4^Lps-del control.

We evaluated systemic splenic cellular immune responses in Leptospira infected and noninfected WT, huTLR4, and muTLR4^Lps-del mice ( Figures 5 and 6 ). Absolute numbers of splenocytes, CD3+, CD3+CD4+, and CD3+CD8+ T cells were quantified ( Figure 5 ). We found that, although not significant, absolute numbers of splenocytes were increased between infected and uninfected groups. Furthermore, between infected groups, absolute numbers of CD3+, CD3+CD4+, and CD3+CD8+ T cells were reduced in huTLR4 in comparison to WT and muTLR4^Lps-del mice.

Further flow cytometric analysis of proportions of splenic T cells from WT, huTLR4 and muTLR4^Lps-del mice infected with L. interrogans ( Figure 6 ) showed no differences in proportions of total CD3+, CD3+CD4+, and CD3+CD8+ T cells between uninfected (PBS) and infected (Inf) mice except for a significantly increased population of CD3+CD4+ and a significantly decreased population of CD3+CD8+ in huTLR4 mice ( Figures 6A–C ). CD4+ naïve T cell populations decreased between all noninfected and infected groups ( Figure 6D ) as they differentiated into effector T cells ( Figure 6E ) but not memory T cells ( Figure 6F ). Thus, as expected, naïve CD4+ T cells differentiated info effector T cells in infected WT, huTLR4 and muTLR4^Lps-del mice. Proportions of CD8+ naïve T cells were slightly lower in infected than in noninfected groups ( Figure 6G ) as they differentiated into CD8+ effector T cells but not memory T cells. Differences in CD8+ effector T cells were not significant except for muTLR4^Lps-del ( Figures 6H, I ).

To get further insight into the role of TLR4 in Leptospira induced inflammation, we compared primary splenocytes from WT and huTLR4 mice, as well as mice with a targeted knock-out mutation in TLR4 (muTLR4^KO) (12). The cells were stimulated with 10⁷/ml and 10⁸/ml heat-killed L. interrogans and gated on the CD11b⁺ macrophage/monocyte population using an ex vivo flow-cytometry based assay (13) ( Figure 7 ). We found that 6% and 20% of CD11b+ macrophages from huTLR4 mice stimulated with 10⁷ and 10⁸ Leptospira, respectively, produced TNF-α in comparison to ~ 14% and 28% from WT, and ~ 3% and 12% from muTLR4^KO. In addition, macrophages from all mouse genotypes responded to the same extent to CpG and to Pam3CSK4 (TLR9 and TLR2 agonists). WT and huTLR4 macrophages responded to LPS from E. coli (EC LPS, TLR4 agonist) in contrast to muTLR4^KO macrophages that did not produce TNF-α. It is important to note that stimulation of WT and huTLR4 macrophages with increasing concentrations of the TLR4 agonist (10ng/ml and 100 ng/ml of EC LPS) led to the production of equivalent amounts of TNF-α. This indicates that the huTLR4 receptor is as functional as the WT, in the murine background. Interestingly, previous quantification of relative TLR4 mRNA expression in bone marrow derived macrophages from WT and huTLR4 mice, using the same primers, showed a significantly higher expression of huTLR4 (18). A lower overall production of TNF-α by huTLR4 macrophages stimulated with 10⁷ and 10⁸ heat-killed Leptospira suggests that the systemic inflammatory response of huTLR4 is reduced compared to WT mice.

The major goal of this study was to understand how TLR4 affects Leptospira infection in vivo. Another goal was to develop an immunocompetent mouse model of leptospirosis. The rationale for using humanized TLR4 transgenic mice is that, if human TLR4 does not sense the atypical leptospiral-LPS (1, 2, 19), these mice should be more permissive of Leptospira escape and more susceptible to leptospirosis than wild-type mice expressing murine TLR4 that do sense leptospiral-LPS. We therefore hypothesized that C57BL/6J mice carrying the human TLR4/MD-2 receptor and co-receptor in its immune cells (huTLR4) should be more sensitive to sublethal leptospirosis than wild-type congenic C57BL/6J (WT). However, we found that dissemination of Leptospira interrogans serovar Copenhageni FioCruz in blood, shedding in urine, colonization of the kidney and overall kinetics of leptospirosis progression is equivalent between mice carrying murine and human TLR4. Furthermore, the inflammatory response to infection with Leptospira seemed to be subdued in huTLR4 compared to WT mice in that we observed less infiltrates of mononuclear lymphocytes, less innate immune markers and no relevant differences in fibrosis markers in kidney as well as less production of TNFα by murine splenic macrophages harboring a fully functional huTLR4 receptor. These results are consistent with the lack of leptospiral-LPS stimulation of human TLR4 (1).

Decreased inflammation in huTLR4 mice could have led to increased numbers of Leptospira in kidney. Surprisingly, this was not observed. We analyzed the kinetics of leptospirosis infection and colonization of the kidney between infected and uninfected humanized TLR4 mice as established previously in our laboratory (8, 10, 20) ( Figures 1 and 2 ). Compared to WT and to muTLR4^Lps-del, infected huTLR4 mice showed a similar pattern of low and slow loss of weight as the WT which was considerably different than the abrupt loss of weight shown by muTLR4^Lps-del at days ~6–7 post-infection when colonization of the kidney takes place, as we previously observed in C3H-HeJ mice (8, 10, 11, 20). The latter carries a point mutation in the toll like receptor 4 (tlr4) cytoplasmic domain that results in a lack of signaling despite presence of the receptor and binding of bacterial LPS (9). Furthermore, dissemination in blood and shedding of Leptospira in urine of WT and huTLR4 mice were equivalent as was the amount of live Leptospira detected in kidney. This again contrasted with the amount of Leptospira detected in blood, urine and kidney of muTLR4^Lps-del mice that were 1Log higher than the WT and huTLR4 counterparts. An even higher load of Leptospira was observed in blood and urine of infected C3H-HeJ mice (8, 20). Furthermore, wild-type C57BL/6J mice were found to have 2Log lower numbers of Leptospira in urine than TLR4^KO mice (4, 21). Interestingly, the results of Leptospira infection obtained in C3H-HeJ (containing a point mutation in the TLR4 receptor) and in C57BL/6J TLR4^KO (do not express the TLR4 receptor) are consistent with the results obtained here with C57BL/6.10J TLR4^Lps-del which contain a spontaneous mutation that completely removes the tlr4 gene and thus results in the absence of the TLR4 receptor. These results confirm that mutations resulting in a loss of response to leptospiral-LPS, either resulting from a point mutation in tlr4 in C3H-HeJ or from a complete deletion of the gene in C57BL6/J mice, increase susceptibility to Leptospira infection regardless of genetic background. However, this was not replicated by substitution of wildtype murine TLR4 with human TLR4, that is known to lack leptospiral-LPS sensing.

Analysis of markers of inflammation and other immune cell engagement ( Figures 3 and 4 ) in kidney showed that huTLR4 mice exhibit less markers of innate immune cell activity (CxCL1/KC, CxCL2/MIP-2 and TNF-α) and less anti-inflammatory cytokine IL-10 than WT mice. Differences in antimicrobial iNOS and fibroblast activation marker collagen 1 A1 (ColA1) which are indicative of interstitial collagen deposition (fibrosis) were not significant between huTLR4 and WT mice. This may be further corroborated by the lower number of mononuclear lymphocyte infiltrates observed in H&E stained slides of the kidney. These data suggest that decreased inflammatory responses in the huTLR4 mouse kidney is dependent on TLR4. If human TLR4 does not sense the atypical Leptospira LPS (1, 2, 19), huTLR4 mice should have been more permissive of Leptospira escape and more susceptible to leptospirosis than wild-type. However, the opposite was observed. The data could suggest that the huTLR4 transgene may not function to their fullest capacity in a mouse background as it does in humans. However, our data from splenic murine macrophages stimulated with Leptospira ( Figure 7 ) clearly shows that the huTLR4 receptor is fully functional in mouse macrophages as they produced as much TNF-α after stimulation with increasing doses of the E. coli LPS TLR4 agonist as did WT mice. Although stimulation of TLR9 and TLR2 led to equivalent production of TNF-α by WT, huTLR4 and muTLR4^Lps-del mouse macrophages, recognition of heat killed Leptospira was decreased in huTLR4 when compared to WT, which is consistent with the lack of sensing of leptospiral-LPS by human TLR4 (1).

When we analyzed proportions of splenic T cells two weeks post infection with L. interrogans ( Figures 5 and 6 ) we found increased populations of CD4+ effector T cells in infected versus uninfected mice in WT, huTLR4 and muTLR4^Lps-del which was consistent with our previous observations in C3H-HeJ mice (8, 10). Furthermore, the differences observed in CD3+CD8+ T cells suggest that cytotoxic CD8+ T cells could be engaged in mounting an immune response to Leptospira. Interestingly, in cattle vaccinated with a Leptospira bacterin and subsequently challenged with L. borgpetersenii serovar Hardjo, CD8+ were the only T cells that proliferated 6 weeks post infection (22). Thus, independently of genetic background, effector CD4+ T cells promote the T cell response to L. interrogans infection and, quite possibly, CD8+ T cells may be involved.

Regarding differences in infection in immunocompetent (C57BL/6J WT) versus immunocompromised mice (C3H-HeJ, and C57BL/6.10J TLR4^Lps-del and C57BL/6J TLR4^KO), Leptospira persistence in WT mice, shown in ours and other studies (21) fulfills their role as host reservoirs. In the current study we show that Leptospira dissemination, although lower in WT C57BL/6J than in C3H-HeJ (8), can be quantified and the immune marker profiles of infection are similar between the two strains. The differences in order of magnitude observed between the two strains simply allow for a more robust quantification of dissemination and inflammation in C3H-HeJ (8, 20). It is however important to note that Leptospira disseminated and caused pathology in wild-type C57BL/6J which carry a fully functional murine TLR4. This could be explained by the recent finding that leptospiral-LPS also subverts murine TLR4 recognition (23). Indeed, leptospiral-LPS recognition by murine TLR4 is only partial, since only the MyD88 pathway is engaged, not the endosomal TRIF pathway. This avoidance is due to LPS-associated lipoproteins that interfere with CD14 binding of leptospiral-LPS, which block the internalization of TLR4. As a consequence of TRIF avoidance, secretion of type I interferons and nitric oxide production upon Leptospira infection are decreased (23). This somewhat altered TLR4 response could explain the dissemination and kidney colonization after Leptospira infection in mice and it can provide a clue about the lack of profound differences between huTLR4 and WT mice after infection. In addition, Leptospira evade different PRRs, such as TLR5 and NOD proteins, which could contribute to the poor efficiency of macrophages and phagocytes facing Leptospira (24). In this context, the overall role of TLR4 sensing Leptospira could be minimal although its presence would still be important. For example, TLR4 expressed by B cells was shown to be key in early IgM production in mice, which is partially able to counteract Leptospira infection (4). We may speculate that early IgM production, activated three days post infection, may be linked to a role of TLR4, as it was recently shown that B1 cell activation and tumor-reactive IgM production were defective in TLR4^KO mice (25).

In conclusion, we expected huTLR4 mice to have been more permissive of Leptospira escape and more susceptible to leptospirosis than congenic wild-type mice. Our hypothesis did not prove to be true. Our data suggests that less recognition of leptospiral-LPS by huTLR4 did not affect overall infection and does not result in increased pathology of the kidney. These data indicate that an intact TLR4 gene, be it mouse or human, is necessary to control pathology caused by Leptospira. Unexpectedly, this could help explain why ~90% of people infected with pathogenic Leptospira (26) are fairly tolerant of this pathogen and do not develop acute Weil’s disease.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

The animal study was reviewed and approved by Institutional Animal Care and Use Committee, University of Tennessee Health Science Center (UTHSC).

MSG is the corresponding author who conceptualized the project, procured funding, designed experiments, data analysis, prepared figures and wrote the manuscript. NN is the first author who performed the experiments, designed experiments, assisted in writing, data analysis, assisted in preparing figures. MS performed experiments. AMH provided HuTLR/MD2 mice, designed and performed experiments, data analysis and critical review. CW provided critical review and assisted in writing the manuscript. All authors contributed to the article and approved the submitted version.

We thank the Flow Cytometry and Cell Sorting Core at UTHSC. This work was supported by Public Health Service grants R44AI096551, R43AI136551, R21AI142129 (MG-S) and from the National Institutes of Health and supported by an Institut Pasteur grant PTR66-2017 (CW).

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.