Leptospiral lipopolysaccharide dampens inflammation through upregulation of autophagy adaptor p62 and NRF2 signaling in macrophages

Leptospira interrogans are pathogenic bacteria responsible for leptospirosis, a worldwide zoonosis. All vertebrates can be infected, and some species like humans are susceptible to the disease whereas rodents such as mice are resistant and become asymptomatic renal carriers. Leptospires are stealth bacteria that are known to escape several immune recognition pathways and resist killing mechanisms. We recently published that leptospires may survive intracellularly and exit macrophages, in part by escaping xenophagy, a pathogen-targeting form of autophagy. Interestingly, autophagy is one of the antimicrobial mechanisms often highjacked by bacteria to evade the host immune response. In this study we therefore explored whether leptospires subvert the key molecular players of autophagy to facilitate the infection. We showed in macrophages that leptospires triggered a specific accumulation of autophagy-adaptor p62 in puncta-like structures, without major alteration of autophagy flux. Unlike active bacterial mechanisms described to date, we demonstrated that leptospires trigger p62 accumulation using a passive mechanism of LPS signaling via TLR4/TLR2. p62 is a central pleiotropic protein, not only involved in autophagy, but also mediating cell stress and death, via the translocation of transcription factors. We demonstrated that Leptospira-driven accumulation of p62 induced the translocation of transcription factor NRF2. However, NRF2 translocation upon Leptospira infection did not result as expected in antioxydant response, but dampened the production of inflammatory mediators such as iNOS/NO, TNF and IL6. Overall, these findings highlight a novel passive bacterial mechanism linked to p62/NRF2 signaling that decreases inflammation and contributes to the stealthiness of leptospires.


INTRODUCTION puncta-like structures 186
To address modulation of autophagy-adaptors upon infection with leptospires, we infected bone marrow 187 derived macrophages (BMDMs) with L. interrogans Manilae strain L495 and analyzed by Western blot 188 the levels of the different adaptors p62, NDP52 and Optineurin. We observed an accumulation of p62 189 over time upon infection (Fig 1A). We confirmed this phenotype using other pathogenic serotypes of 190 L. interrogans: Copenhageni strain Fiocruz L1-130 and Icterohaemorrhagiae strain Verdun, and using 191 the saprophytic Leptospira biflexa Patoc strain Patoc I (Fig 1B). Epifluorescence analyses of BMDMs 192 infected 4h with L. interrogans further revealed that infection triggers accumulation of p62 in puncta-193 like structures (Fig 1C). This phenotype seemed specific of p62 since we did not observe accumulation 194 of the autophagy adaptors (NDP52 & optineurin) upon leptospira infection (Sup . Fig 2A). 195 Among other mechanisms, autophagy adaptors are constitutively degraded by autophagy upon 196 autophagosome / lysosome fusion. We hypothesized that p62 accumulation could be triggered by 197 blockage of autophagy flux by leptospires. However, when we monitored the autophagy hallmark 198 protein LC3-II in BMDMs by Western blot, no LC3-II accumulation was observed upon Leptospira 199 infection, in contrast to Bafilomycin (BafA1) treatment that blocks the autophagy flux, causing LC3-II 200 to accumulate (Sup. Fig 1A). Thus we conclude that infection with leptospires does not alter the 201 autophagy flux. Unexpectedly, we observed a mild reduction in LC3-II levels upon infection 202 (Sup. Fig 1B). Using murine macrophages RAW-mLC3-diFluo cells analyzed by automated 203 microscopy, we were able to confirm that leptospires induce a mild decrease of the number of 204 autophagosomes without altering the number of autolysosomes, again showing no alteration of the 205 autophagy flux (Sup. Fig 1C-D). 206 We then analyzed the transcriptional regulation of p62 in BMDMs by RT-qPCR and observed a 207 significant upregulation of p62 mRNA 24 h post-infection with L. interrogans (Fig 1D). This 208 corroborated the idea that p62 accumulation is not mediated by autophagy blockage. The kinetics of the 209 p62 puncta formation were further characterized in RAW264.7 cells, using automated high content (HC) 210 confocal microscopy. Results showed time-dependent increase of both the number of p62 puncta per 211 cell and the percentage of p62 positive cells (Fig 1E & Sup. Fig 2B). Single cell analysis confirmed an 212 average of 5-10 puncta per cell and highlighted cell-to-cell heterogeneity, with some macrophages 213 containing up to 60 puncta (Fig 1F). Of note, BafA1 that blocks autophagy flux also led to p62 214 accumulation as expected, but to a much lower extent than infection with leptospires (Fig 1C, 1E & 1F). 215 Altogether, these results confirm our previous results showing that L. interrogans do not induce 216 autophagy in murine macrophages (14), and suggest that leptospires induce a specific accumulation of 217 p62, not linked to an autophagy blockage. 218

p62 accumulation is triggered by the leptospiral LPS through TLR4 & TLR2 219
Many bacteria interfere actively with autophagy molecules via secreted effectors or RNA 220 interference (16, 23-28). To investigate whether such active mechanisms are also used by leptospires, 221 we analyzed BMDMs and RAW264.7 cells after stimulation with heat-killed (HK) leptospires. We 222 observed p62 accumulation visible by Western blot (Fig 2A) and quantified by automated microscopy 223 (Fig 2B & 2C). These findings exclude an active leptospiral mechanism and suggest a role for the 224 recognition of leptospiral MAMPs. We therefore stimulated cells for 24h with leptospiral LPS and 225 observed that such stimulation recapitulates p62 accumulation (Fig 2A, 2B & 2C). As p62 accumulation 226 seemed to be mediated by recognition of the leptospiral LPS, we investigated the roles of TLR4 & 227 TLR2. In murine cells, these two receptors are respectively activated by the leptospiral lipid A, and the 228 leptospiral lipoproteins that co-purify with the LPS (5, 12). We infected WT and TLR2/4 dko BMDMs 229 with L. interrogans for 24h, and analyzed by Western blot and automated microsopy. We observed that 230 p62 accumulation was greatly dampened in TLR2/TLR4 dko cells (Fig 2D & 2E). Finally, we analyzed 231 mRNA levels by RT-qPCR and observed that the increase in p62 mRNA levels observed in WT 232 BMDMs was abolished in TLR2/4 dko BMDMs (Fig 2F). Overall, these data show that p62 233 accumulation is mediated by TLR2/4 recognition of the leptospiral LPS. To address if such mechanism 234 was conserved in vivo, we injected C57BL/6 mice intraperitoneally with 1 x10 8 heat-killed of not replicating nor disseminating upon injection. Interestingly, using automated microscopy we 237 showed that p62 was also accumulated in vivo in peritoneal F4/80 + macrophages (Fig 2G). 238 Finally, we ask whether this mechanism of p62 accumulation could be conserved in human 239 cells. Indeed, there is a TLR4 host species-specificity of the innate immune recognition of leptospires 240 (29). Leptospiral lipid A activates mouse-TLR4 but not human-TLR4 (12). However, lipoproteins that 241 co-purify with the LPS activate both mouse-TLR2 and human-TLR2 in a CD14-dependent manner (12). 242 We therefore infected human monocytic THP1-CD14 cells with L. interrogans serovar Manilae strain 243 L495. We also observed a specific p62 accumulation and no NDP52 accumulation after infection with 244 leptospires (Sup. Fig 2C), suggesting conserved mechanism of p62 accumulation in human and murine 245 macrophages, and a prominent role of TLR2 in human cells. 246 Leptospires have been shown to activate the NLRP3 inflammasome, in humans (30, 31), and in 247 a TLR2/4-dependent manner in mice (7). Furthermore, NLRP inflammasomes have been shown to 248 modulate autophagy (32-35). Therefore, we asked whether activation of inflammasome could play a 249 role in p62 accumulation and LC3-II diminution. We stimulated BMDMs with both live and heat-killed 250 leptospires in the presence or absence of NLRP3 inhibitor glibenclamide. Efficiency of the treatment 251 was controlled by measuring IL1 production after 24h (Sup. Fig 3A). p62 and LC3-II levels were 252 analyzed by Western blot. We observed a similar levels of p62 accumulation upon infection with either 253 live or heat-killed leptospires in both control and glibenclamide treated cells (Sup. Fig 3B). 254 Consistently, the diminution of LC3-II upon infection was visible in both control and glibenclamide 255 treated cells (Sup. Fig 3B). Overall, this suggests that the modulation of autophagy players by 256 leptospires, although TLR2/4-dependent, is not mediated by activation of the NLRP3 inflammasome. 257

Leptospires and their LPS trigger translocation of transcription factor NRF2 258
Stress pathways are induced when autophagy adaptors accumulate in the cell (i.e. in the absence of 259 functional autophagy or because of specific upregulation). p62 accumulation induces translocation of 260 stress-responsive nuclear factor erythroid 2-related factor 2 (NRF2) (20, 36, 37). Subsequently, NRF2 261 triggers antioxidant and antiapoptotic programs (20, 36), and promotes p62 upregulation, creating a loop that counteracts stress in autophagy-deficient conditions (36). We therefore infected BMDMs with 263 L. interrogans serovar Manilae strain L495 to analyze NRF2 by immunofluorescence 4h post-infection. 264 Whilst NRF2 staining was barely visible in the non-infected conditions, NRF2 staining in cell nuclei 265 was clearly evident upon infection (Fig 3A). The intensity of NRF2 staining in the nucleus peaked 266 around 4h post-infection (Fig 3B). We next performed single cell analysis of NRF2 fluorescence 267 intensity in the nucleus of RAW264.7 cells infected for 4h with three different serovars of 268 L. interrogans, and observed similar NRF2 increase for all the strains (Fig 3C, left panel). Next, we 269 calculated the ratio of NRF2 intensities [nucleus/cytoplasm] and estimated that cells positive for NRF2 270 had a ratio > 1.4. We then plotted the percentage of cells with NRF2 ratio superior to 1.4 and observed 271 that up to 60% of cells were positive after 4h of infection (Fig 3C, right panel). Finally, to understand 272 the contribution of leptospiral LPS in triggering NRF2, we performed similar analyses on BMDMs 273 stimulated for 4h with either live, heat-killed leptospires or their purified LPS. All conditions triggered 274 similar NRF2 translocation (Fig 3D), suggesting that the leptospiral LPS, that triggers accumulation of 275 p62, is responsible for NRF2 activation. Consequent to p62 accumulation and NRF2 translocation upon infection with L. interrogans, we 278 investigated the potential regulation mechanisms between these two phenotypes. We performed 279 small-interfering RNA (siRNA) transfection in BMDMs to knock-down specifically p62 or NRF2, and 280 analyzed NRF2 and p62 4h post-infection with leptospires by automated microscopy. We observed that 281 the NRF2 ratio [nucleus/cytoplasm] was lower upon infection in the p62 siRNA condition (Fig 4A, left  282 panel). Conversely, p62 puncta formation was reduced upon infection in NRF2 siRNA transfected 283 BMDMs (Fig 4B, left panel). Both p62 and NRF2 knocked-down were confirmed and quantified by 284 automated microscopy (Fig 4A & 4B, right panels). Overall, these results showed that p62 contributed 285 to NRF2 translocation, and that, in turn, NRF2 regulated p62 puncta formation, suggesting that p62 and 286

NRF2 are in a feedback loop upon infection by L. interrogans.
To further characterize the involvement of NRF2 in p62 activation, we analyzed BMDMs 288 transfected with NRF2 siRNA and infected with L. interrogans by Western blot and RT-qPCR 24h 289 post-infection, allowing us to measure both the protein and mRNA levels of p62. Interestingly, we 290 observed that NRF2 silencing prevented p62 protein accumulation (Fig 4C) but also reduced mRNA 291 upregulation (Fig 4D), suggesting that NRF2 is a transcriptional regulator of p62 activation. 292

NRF2 translocation prevents inflammation upon infection 293
NRF2 is a stress response transcription factor that promotes antioxidant and antiapoptotic programs 294 upon activation (20, 36). Therefore, we monitored the transcription of targets genes involved in fighting 295 oxidative stress in macrophages in response to infection. We infected BMDMs with L. interrogans and 296 NRF2 has also been shown to have repressor functions and to dampen inflammation by 301 inhibiting the transcription of cytokines (38, 39), hence conferring resistance to inflammatory disease 302 such as sepsis (38). We therefore investigated the role of NRF2 translocation in inflammation upon 303 infection by leptospires. We transfected RAW264.7 cells with siRNA targeted against NRF2 and then 304 infected them with L. interrogans. As expected, considering that NRF2 is known to prevent cell death, 305 we observed a decrease in macrophage viability only upon infection of NRF2-silenced cells 306 (Fig 5A, left panel). Interestingly, this enhanced loss of viability was not associated with lactate 307 dehydrogenase (LDH) release from the cytosol, illustrating that no membrane damage or cell lysis 308 occurred (Fig 5A, right panel). We then analyzed the mRNA levels of several cytokines and enzymes, 309 namely iNOS (nitric oxide inducible synthase), IL6, TNF and IL10. Upon infection, all mRNA levels 310 were higher in NRF2-silenced cells than in control cells (Fig 5B), suggesting that NRF2 does play a role 311 in repressing the expression of inflammatory targets. Finally, to address whether this transcriptional 312 regulation was strong enough to alter cytokines levels, we analyzed nitric oxide (NO), IL6, TNF and IL10 production in cell supernatant. Consistent with the mRNA analyses, we observed higher NO and 314 TNF upon infection of NRF2-silenced RAW264.7 (Fig 5C). These findings were even more striking 315 after normalizing cytokine production against cell viability, as measured using MTT 316 (Fig 5C, grey areas). Increased levels of IL6 were observed after normalizing for cell viability, whereas 317 IL10 was not increased under any condition (Fig 5C, grey areas). Overall, our results show that NRF2 318 plays a repressor role to dampen production of inflammatory mediators such as NO, TNF and IL6 in 319 we conclude that the leptospires behave differently from other pathogenic bacteria that actively 337 modulate autophagy, and for which p62 accumulates on the surface of intracellular bacteria.
We further demonstrated that p62 accumulation was induced by leptospiral LPS signaling 339 through TLR2/TLR4. The unique leptospiral LPS, the best characterized virulence factor (10), plays an 340 important role in leptospiral host-pathogen interactions. We previously showed that leptospiral LPS 341 avoids human TLR4 and mouse TLR4-TRIF activation (13). Recently, leptospiral LPS has also been 342 shown to prevent cell death by pyroptosis (43). Therefore, it was not surprising to find the leptospiral 343 LPS responsible for the p62-puncta phenotype. Interestingly, the LPS of L. biflexa has been shown to 344 be a more potent TLR4 agonist than the LPS of pathogenic L. interrogans (13) few negative regulators, and it is hypothesized that its main regulator would be autophagy, through 359 degradation of p62 (36). Therefore, we hypothesize that since infection with leptospires does not trigger 360 autophagy, the feedback loop between p62 and NRF2 remains active, explaining why p62 puncta did 361 not disappear even at 24h post-infection. Overall, out data indicate that leptospires are potent activators 362 of the p62/NRF2 axis in macrophages. 363 Interestingly, NRF2 has also been shown to have transcription inhibition properties. In 364 BMDMs, NRF2 was described as an inhibitor of transcription for pro-inflammatory targets such as IL6, IL1 and IL12 (39, 46). Our results showed that this is also the case upon infection with L. interrogans. 366 NRF2 inhibits polymerase III recruitment and hence prevents transcription of cytokines in response to 367 stimulation (39). Consequently, NRF2 was shown to downregulate neutrophils activation and 368 migration (47). Interestingly, although neutrophils are abundant in the blood of mice and humans 369 infected with leptospires, they are barely observed in the kidneys, the niche occupied by leptospires 370 during chronic infection (48, 49). Whether Leptospira modulate NRF2 translocation in neutrophils via 371 their LPS to favor survival in the kidneys remains to be studied. In addition, we showed that the 372 accumulation of p62 observed in murine macrophages was conserved in human cells. Interestingly, 373 human cells do not sense leptospiral lipid A through TLR4 (12), leading us to hypothesize that TLR2 374 alone could be responsible for sensing leptospires and inducing p62 accumulation in THP1 cells. 375 Although we could not address NRF2 translocation in human cells due to lack of specific tools, we 376 speculate that inflammation dampening might be conserved in human macrophages. 377 Other microbes such as Epstein-Barr virus (EBV) or parasite Leshmania major have been 378 shown to activate the p62/NRF2 axis (50, 51), hence showing an important role of NRF2 in response to 379 pathogens. However, to date, the role of NRF2 translocation in response to pathogens remains unclear. 380 Among others, NRF2 is involved in the induction of antioxidant program upon infection (36). We were 381 therefore surprised to observe no modulation in the mRNA levels of antioxidant targets upon infection. 382 Of note, pathogenic leptospires are equipped to fight against oxidative stress with inducible catalase, a 383 virulent factor, peroxidase, and peroxiredoxin (52). Interestingly, active repression of these 384

NRF2-dependent antioxidant targets was previously shown upon infection with live Leshmania major 385
parasite (51, 53). Whether leptospires could also actively prevent upregulation of NRF2 antioxidant 386 targets remains to be addressed. 387 In summary, we have demonstrated that leptospires passively subvert the p62-NRF2 axis 388 through LPS activation of TLR signaling and this leads to a reduction of inflammatory mediators. This