Leptospiral LPS escapes mouse TLR4 internalization and TRIF-associated antimicrobial responses through O antigen and associated lipoproteins

Leptospirosis is a worldwide re-emerging zoonosis caused by pathogenic Leptospira spp. All vertebrate species can be infected; humans are sensitive hosts whereas other species, such as rodents, may become long-term renal carrier reservoirs. Upon infection, innate immune responses are initiated by recognition of Microbial Associated Molecular Patterns (MAMPs) by Pattern Recognition Receptors (PRRs). Among MAMPs, the lipopolysaccharide (LPS) is recognized by the Toll-Like-Receptor 4 (TLR4) and activates both the MyD88-dependent pathway at the plasma membrane and the TRIF-dependent pathway after TLR4 internalization. We previously showed that leptospiral LPS is not recognized by the human TLR4, whereas it signals through murine TLR4, which mediates mouse resistance to acute leptospirosis. However, leptospiral LPS has low endotoxicity in mouse cells and is an agonist of TLR2, the sensor for bacterial lipoproteins. Here, using confocal microscopy and flow cytometry, we showed that the LPS of L. interrogans did not induce internalization of TLR4 in mouse macrophages, unlike the LPS of Escherichia coli. Consequently, the LPS failed to induce the production of the TRIF-dependent nitric oxide and RANTES, both important antimicrobial responses. Using shorter O antigen LPS and repurified leptospiral LPS with reporter HEK cells, we further found this TLR4-TRIF escape to be dependent on both the co-purifying lipoproteins and the full-length O antigen. Furthermore, our data suggest that the O antigen could alter the binding of the leptospiral LPS to the co-receptor CD14 that is essential for TLR4-TRIF activation. Overall, we describe here a novel immune escape mechanism linked to leptospiral LPS. We hypothesize that the LPS, already known as a virulence factor, plays a major role in the innate immune evasion of the leptospires, thereby contributing to their stealthiness and chronicity in mice. Author summary Leptospira interrogans is a bacterial pathogen, responsible for leptospirosis, a worldwide neglected reemerging disease. L. interrogans may cause an acute severe disease in humans, whereas rodents and other animals asymptomatically carry the leptospires in their kidneys. They can therefore excrete live bacteria in urine and contaminate the environment. Leptospires are stealth pathogens known to escape the innate immune defenses of their hosts. They are covered in lipopolysaccharide (LPS), a bacterial motif recognized in mammals through the Toll-like receptor 4 (TLR4), which triggers two different signaling pathways. We showed previously that pathogenic leptospires escape TLR4 recognition in humans. Here we show in mice that the leptospiral LPS triggers only one arm of the TLR4 pathway and escapes the other, hence avoiding production of antimicrobial compounds. Removing the lipoproteins that always co-purify with the leptospiral LPS, or using shorter LPS, restores the stimulation of both pathways. This suggests a novel escape mechanism linked to the LPS and involving lipoproteins that could be instrumental for leptospires to escape the mouse defense and allows for their chronic renal colonization.


Author summary 38
Leptospira interrogans is a bacterial pathogen, responsible for leptospirosis, a worldwide neglected 39 reemerging disease. L. interrogans may cause an acute severe disease in humans, whereas rodents and 40 other animals asymptomatically carry the leptospires in their kidneys. They can therefore excrete live 41 bacteria in urine and contaminate the environment. Leptospires are stealth pathogens known to escape Treponema pallidum, respective agents of Lyme disease and syphilis, Leptospira spp. possess a Toll-Like-Receptor 4 (TLR4) and Toll-Like-Receptor 2 (TLR2) sense bacterial LPS and bacterial hand, Loa22, another abundant lipoprotein was shown to be essential for leptospiral virulence [32].

121
Even though most of the functions of these lipoproteins remain a mystery, previous publications 122 showed that some leptospiral lipoproteins play a role in the innate immune escape. Indeed, LipL21 123 binds tightly to the leptospiral peptidoglycan and hence prevents its recognition by the PRRs NOD1 124 and NOD2 [33]. Additionally, LipL21 and other lipoproteins have also been described as inhibitors of 125 neutrophil myeloperoxidase [34].

126
To our knowledge, the mechanism underlying the low endotoxicity of the leptospiral LPS seen in 127 mouse cells [26,27,35] has never been investigated. Moreover, the role of the co-purifying lipoproteins 128 that often contaminate leptospiral LPS preparations and confer their TLR2 activity remains to be 129 studied. Hence, in this study we aimed to characterize the atypical signaling of the leptospiral LPS in 130 murine cells and to elucidate the role of these co-purifying lipoproteins. 178 Surprisingly, we showed that at lower concentration, leptospiral LPS avoided not only the 179 TRIF-dependent NO production, but also the production of MyD88-dependent KC ( Figure S2C).

180
These results suggest that at high concentrations, the leptospiral LPS escapes the TLR4 internalization 181 and specific TRIF-dependent responses, whereas at low concentrations, the leptospiral LPS escapes 182 both pathways.

185
Since at high concentration of leptospiral LPS, the stimulation of TLR4-MyD88 responses was 186 triggered, we hypothesized that the phenotype we observed was not due to the interaction of lipid A 187 with TLR4, but that the O antigen of the LPS could be involved in the TRIF escape. Therefore, we   Figure 3C). Furthermore, we showed that this shorter LPS induced more In addition, we also purified and characterized the LPS of the saprophytic L. biflexa serovar Patoc 201 strain Patoc and showed that it was much shorter than the LPS of the pathogenic species ( Figure S3A).

202
Because the structure of the Patoc Lipid A is unknown, we could not estimate its exact molar 203 concentration. Nevertheless, based on same LPS amount by weight, HEK reporter system analyses 204 showed that both Patoc and Verdun LPS had comparable mouse TLR4 activity ( Figure S3B), whereas 205 the LPS of Patoc had much lower TLR2 activity than the Verdun LPS ( Figure S3B). We further 206 confirmed by analyses of TLR4 KO and TLR2 KO BMDMs that the signaling of Patoc LPS was 207 TLR2-independent ( Figure S3C). Consistent with the results of the M895 shorter LPS, we showed that 208 the short Patoc LPS induced in macrophages a higher production of RANTES and NO than the 209 Verdun LPS ( Figure S3D). Altogether these data strongly suggest that the full-length intact LPS is 210 important to avoid TLR4 internalization and subsequent TRIF-dependent responses.

213
After investigating the bacterial components that could be involved, we decided to assess whether host 214 factors could also be implicated in the mechanism. We first investigated the role of serum in the 215 signaling. Both full-length and shorter leptospiral LPS benefited from serum addition for RANTES 216 production ( Figure S4). Next, we investigated whether serum components could contribute to the 217 differential signaling observed between the full-length and the shorter LPS. CD14 appeared as a good 218 candidate given its role in TLR4 endocytosis [25] and in the signaling of leptospiral LPS via TLR2 219 [27]. Stimulation of RAW264.7 macrophages in serum-free medium, supplemented with increasing 220 doses of recombinant soluble CD14 (sCD14) revealed that the NO production in response to the 221 shorter M895 LPS was dependent on the presence of sCD14 ( Figure 4A). Surprisingly, the presence of 222 sCD14, even at 500 ng/mL, did not trigger the production of NO by the full-length L495 leptospiral 223 LPS ( Figure 4A). sCD14 mediates TLR4-signaling by binding and delivering LPS to the TLR4-MD2 224 complex [22]. Thus, to further characterize the role of sCD14 in the M895 LPS signaling, binding 225 assays were performed on high resolution 20 % acrylamide gels and visualized by silver staining. The 226 band corresponding to sCD14 shifted in the presence of both LPS (that remains in the upper part of the gel). However, the shift and disappearance of the sCD14 band was much more prominent with 228 0.01 pmol of the shorter M895 LPS than with the same dose of the L495 LPS ( Figure 4B, upper panel).

229
These results were confirmed by WB targeting sCD14 via the polyHistidine tag ( Figure 4B, 230 lower panel). Overall, these results suggest that the full-length LPS has a lower affinity for sCD14 231 than the shorter LPS, which could explain why it escapes TLR4 internalization and TRIF-responses.  Figure 5B). Furthermore, silver staining analysis showed that the overall structure of the LPS 245 was not altered by the treatment. Among other lipoproteins, we identified the LipL21 ( Figure 5C) and 246 LipL32 (data not shown) as co-purifying with the LPS; both were significantly decreased upon 247 repurification. Stimulation of macrophages by the original and repurified LPS showed that the 248 localization of TLR4 was slightly modified when stimulated with the repurified LPS compared with 249 the original LPS ( Figure 5D and 5E). In BMDMs we observed consistently increased production of 250 TRIF-dependent RANTES and NO, but not of MyD88-dependent KC upon stimulation with the 251 repurified LPS ( Figure 5F). These findings were confirmed by mRNA analyses of RANTES and iNOS ( Figure 5H). These results suggest that the presence of co-puryfing lipoproteins would have an 254 inhibitory role and would impair the leptospiral signaling through TRIF.

255
To confirm the implication of the co-purifying lipoproteins in escape from the TLR4-TRIF pathway, 256 we also treated the leptospiral LPS preparations with either proteinase K or lipase. Results with the 257 HEK reporter system showed that these treatments did not affect mTLR4 activity. Interestingly, lipase 258 digestion only (not proteinase K) resulted in the expected reduction in the TLR2 activity of the LPS 259 ( Figure S5A and S5B). Consistent with the repurification data, leptospiral LPS preparations digested 260 with proteinase K (and not lipase) induced increased RANTES production in macrophages 261 ( Figure S5C and S5D). We checked that the proteinase K treatment did not alter the structure of the 262 leptospiral LPS ( Figure S5E) and that it efficiently degraded flagellin ( Figure S5F). All together, these 263 data suggest that the lipoproteins contribute to the TLR4-TRIF escape via the protein moiety and not 264 via the lipid anchors. Since LipL32 is the major lipoprotein present in the outer membrane of 265 leptospires and is the only one to have been shown to be a TLR2 agonist [27,31], we also purified and 266 stimulated macrophages with the LPS of L. interrogans Manilae mutant strain LipL32. Although the 267 molar concentration of the LPS preparations was not measured, using similar amounts (as estimated 268 by weight), the mutant LPS did not have a lower TLR2 activity than the WT LPS ( Figure S5G).

269
Additionally, there was no increased production of RANTES or NO ( Figure S5H). These results

270
suggest that the phenotype was not due to a single specific lipoprotein. This led us hypothesize that 271 lipoproteins are redundant regarding escape of the internalization of TLR4.

274
Our data showed that leptospiral LPS escapes TLR4 internalization and that full-length LPS and co-

301
Conversely, we hypothesize that the poor binding of the leptospiral LPS to sCD14 could limit 302 TLR4-TRIF signaling. Furthermore, we showed that at lower concentrations, leptospiral LPS escapes 303 both the MyD88 and the TRIF pathways, as illustrated by the lack of KC and NO production. Again, 304 the poor binding of the LPS to CD14 is consistent with these results, given that CD14 is essential for

325
In addition, we showed that the copurifying and omnipresent lipoproteins were also responsible, at 326 least in part, for the lack of TLR4 intracellular signaling. We hypothesize that these lipoproteins also 327 affect CD14 binding and that they could potentially play a role in CD14 sequestration. Indeed, CD14

406
Overall, these studies suggest that TRIF plays a moderate role in the outcome of the infection, unlike

414
To conclude, the recognition by TLR4 is essential for host resistance against leptospirosis. Although 415 leptospiral LPS is recognized by the mTLR4, we described here that in vitro it escapes the TRIF arm      Table S1. RT-qPCR was performed on 524 a StepOnePlus Real Time PCR System (Applied Biosystems) with the standard protocol (2 h) and fold 525 change was calculated with the 2 -C method on the average of the technical duplicates.

528
Cells were seeded at 0,2 x 10 6 cells/mL in 24-well plates containing 12 mm round coverslips (#1.5 529 mm thickness, Electron Microscopy Science) with 1 mL per well and stimulated as described above.

543
Argon for AF488 and 594HeNe for AF594, with power between 5 to 15%. Sequential acquisition by 544 channel was performed using a hybrid detector for AF488 (gain range from 70 % to 100 %) and