Clostridium difficile Toxin B activates the NLRP3 inflammasome in human macrophages, demonstrating a novel regulatory mechanism for the Pyrin inflammasome

Pyrin is a cytosolic immune sensor that forms an inflammasome when bacterial virulence factors inhibit RhoA, triggering the release of inflammatory cytokines, including IL-1β. Gain of function mutations in the MEFV gene encoding Pyrin cause auto-inflammatory disorders, such as familial Mediterranean fever (FMF) and Pyrin associated auto-inflammation with Neutrophilic Dermatosis (PAAND). To precisely define the role of Pyrin in detecting pathogen virulence factors in relevant human immune cells, we investigated how the Pyrin inflammasome response was initiated and regulated in monocyte-derived macrophages (hMDM) compared to human monocytes. Unlike monocytes and murine macrophages, we determined that hMDM failed to activate Pyrin in response to known Pyrin activators Clostridioides difficile (C. difficile) toxins A or B (TcdA or TcdB). In contrast, TcdB activated the NLRP3 inflammasome in hMDM. Notably, we ascertained that the Pyrin inflammasome response could be re-enabled in hMDM by prolonged priming with either LPS or type I or II interferons, and required an increase in Pyrin expression. These data demonstrate an unexpected redundancy in detecting these toxins by inflammasome sensors.

pseudomembranous colitis. All of these effects are entirely dependent on the 126 expression of TcdA and TcdB, although TcdB is sufficient to cause disease [16]. PBMCs, monocytes, and neutrophils [11]. 139 140 In this study, we assessed the inflammasome response to C. difficile and its 141 toxins, TcdA and TcdB, in a human macrophage model using M-CSF 142 monocyte-derived macrophages. We discovered that only TcdB triggered an 143 inflammasome response in these cells and that, in contrast to monocytes, this 144 was entirely independent of Pyrin and instead occurred through NLRP3. 145 Furthermore, we found that prolonged exposure to LPS or type-I or -II 146 interferons was sufficient to reactivate the Pyrin inflammasome by increasing 147 Pyrin expression. These results demonstrate that Pyrin is held in an inactive 148 state when monocytes differentiate to macrophages by a hitherto 149 uncharacterized regulatory mechanism. 150 . CC-BY 4.0 International license made available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprint this version posted July 7, 2021. expression when the cells were infected directly (Fig. S1a). Notably, 174 which is transcribed independent of TLR stimulation [21], was still secreted 175 . CC-BY 4.0 International license made available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprint this version posted July 7, 2021. CC-BY 4.0 International license made available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprint this version posted July 7, 2021. ; https://doi.org/10.1101/2021.07.07.451422 doi: bioRxiv preprint

C. difficile toxin B, but not toxin A, mediates NLRP3-dependent 200 inflammasome activation in hMDM 201 202
We observered that hMDM released IL-1β and IL-18 only in response to the 203 toxin-proficient bacteria. To determine whether one or both toxins could trigger 204 an inflammasome response, we incubated the cells with either recombinant 205 toxin A (TcdA) or toxin B (TcdB). As a control, we incubated monocytes with 206 both toxins, as they have previously been shown to respond to both in a Pyrin 207 inflammasome-dependent manner. Monocytes, as predicted, released IL-1β in 208 response to both TcdA and TcdB (Fig. 1b). Surprisingly, and in contrast to the 209 monocytes, we found that TcdB, but not TcdA, induced an inflammasome 210 response in the hMDM (Fig. 1b). Notably, TcdB triggered IL-1β release from 211 hMDM at concentrations as low as 1 ng/ml. To ensure that the lack of TcdA-212 mediated inflammasome activation was not due to a failure of toxin uptake, we 213 assessed TcdA mediated Rac modification. To do so, we used a previously 214 described monoclonal antibody that no longer recognizes Rac when its epitope 215 is modified by the toxin [22]. The antibody was unable to bind to Rac in both 216 monocytes and hMDM treated with the active forms of TcdB and TcdA, but not 217 with mutants lacking glucosyltransferase activity, indicating that the cells took 218 up the toxin (Fig. S1b). 219 220 Given the disparity in inflammasome response between monocytes and hMDM, 221 we assessed whether Pyrin was differentially expressed in the two cell types. 222 Pyrin levels were comparable in hMDM and monocytes (Fig. 1c), indicating that 223 Pyrin expression was not limiting. 224 . CC-BY 4.0 International license made available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprint this version posted July 7, 2021. TcdB to the hMDM. We found that WT BLaER1 cells released IL-1β in response 257 to treatment with TcdB, as well as activating caspase-1 as measured by a 258 caspase-1 activity assay (Fig. 2a). These data demonstrate that TcdB activates 259 an inflammasome response in BLaER1 similar to hMDM. (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprint this version posted July 7, 2021. Accordingly, we generated a BLaER1 cell line overexpressing ASC-mCherry 290 and stimulated it with either TcdB or Nigericin in the presence or absence of 291 CP-456,773. This experiment was also performed in the presence of the 292 caspase-1 inhibitor VX-765 to prevent cell death of inflammasome-activated 293 cells. Following stimulation, the cells were fixed, and the number of ASC specks 294 was quantified by microscopy, followed by normalization to the number of nuclei 295 in each image. Consistent with our other data, we found that TcdB and Nigericin 296 also caused CP456,773-sensitive ASC speck formation in these cells (Fig. 2b). 297 298 . CC-BY 4.0 International license made available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprint this version posted July 7, 2021. ; https://doi.org/10.1101/2021.07.07.451422 doi: bioRxiv preprint To determine the sensor responsible for the TcdB mediated inflammasome 299 response, we used either NLRP3 or Pyrin knock-out BLaER1 cells. As the 300 NLRP3 KO cells we used were on a caspase-4 KO background, we first 301 ensured that caspase-4 played no role in the response. Accordingly, we tested 302 the inflammasome response to TcdB in the caspase-4 KO BLaER1 cells, but 303 found no difference in IL-1β secretion, demonstrating caspase-4 was not 304 required for the response to TcdB (Fig. S2a). The NLRP3, caspase-4 double 305 KO cells were then reconstituted with either NLRP3, the inactive Walker A/B 306 NLRP3 mutant, or a vector alone control. NLRP3 expression was confirmed by 307 immunoblot (Fig. 2c). As previously, we primed these cells with LPS, incubated 308 them with either TcdB, Nigericin or needletox, and assessed IL-1β release to 309 the supernatant. In agreement with our findings in hMDM, only the cells 310 reconstituted with active NLRP3, but not the walker A/B mutant, were able to 311 respond to TcdB and Nigericin (Fig. 2d). In contrast, all cell lines responded 312 equally to the NLRC4 trigger needletox (Fig. 2d) and secreted similar levels of 313 TNFa in response to LPS (Fig. 2d). These results confirm that the TcdB-314 mediated inflammasome response was dependent on the expression of active 315

NLRP3. 316 317
To confirm that Pyrin was not required for the TcdB-driven inflammasome 318 response, we reconstituted Pyrin KO BLaER1 cells with either Pyrin or vector 319 control, and confirmed expression levels by immunoblot (Fig. 2e). We found the 320 inflammasome response to TcdB was unaffected by the absence of Pyrin, 321 indicating that Pyrin does not play a role in the inflammasome response to TcdB 322 . CC-BY 4.0 International license made available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprint this version posted July 7, 2021. ; https://doi.org/10.1101/2021.07.07.451422 doi: bioRxiv preprint in these cells (Fig. 2f). Similarly, Pyrin expression did not affect the 323 inflammasome response to Nigericin or needletox (Fig. 2f). (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprint this version posted July 7, 2021. ; https://doi.org/10.1101/2021.07.07.451422 doi: bioRxiv preprint (Fig. 2g). As anticipated, this response also required ASC and caspase-1, but 348 not Pyrin. Nigericin similarly was NLRP3-dependent, while the NLRC4 trigger 349 only required ASC and caspase-1, demonstrating the effect of NLRP3 ablation 350 was specific (Fig. 2g). Furthermore, the different KO lines secreted similar 351 amounts of TNFa in response to LPS (Supp. Fig 2d) (Fig. 3a). This was also true in peritoneal macrophages, which 382 showed little difference in toxin-mediated IL-1β secretion between WT and 383 NLRP3 KO cells (Fig. 3b). In contrast, the inflammasome response to Nigericin 384 was ablated entirely in the NLRP3 KO. At the same time, there was no 385 difference in IL-1β release in response to transfected dA:dT and no difference 386 in TNFa secretion in response to LPS. We also assessed IL-1β and caspase-1 387 cleavage in WT and NLRP3 KO BMDM following LPS priming and stimulation 388 with TcdB, Nigericin or dA:dT. Similarly, we found no differences between the 389 two genotypes when stimulated with TcdB or the specificity control dA:dT, while 390 Nigericin mediated IL-1β and caspase-1 cleavage were ablated in the NLRP3 391 KO (Fig. 3c). LPS (10ng/ml), TNFa (50ng/ml), IL-10 (100ng/ml), IFN-β (5000 U/ml), IL-4 444 (1000U/ml) or Pam3Cys4K (20ng/ml) for either 5 or 18h. Representative of 3 445 . CC-BY 4.0 International license made available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprint this version posted July 7, 2021.  We investigated this by stimulating the cells with either LPS or Pam3 for 12h, 453 then assessing mRNA copy number for Pyrin by qPCR, using IL-1β as a control. 454 We observed that LPS, but not Pam3CS4K, caused an increase in MEFV 455 transcript compared to the untreated cells (Fig. 4b). In contrast, both LPS and 456 Pam3CS4K increased IL-1β transcription, demonstrating that the increase in 457 MEFV transcript was specific to LPS (Fig. 4b). It is, therefore, likely that the 458 increase in Pyrin expression is driven by increased gene transcription. 459

LPS and interferons prime activation of the Pyrin inflammasome in hMDM 461 462
We tested whether increased Pyrin expression would be sufficient to enable 463 Pyrin inflammasome activation. Accordingly, we primed hMDM with LPS for 464 either 3 or 18 hours, pre-incubated them with DMSO, CP-456,773, VX-765 or 465 colchicine, and then treated them with TcdA or Nigericin. As we had observed 466 previously, TcdA did not trigger an inflammasome response after 3h of LPS 467 priming (Fig. 5a). In contrast, after 18 h, TcdA triggered robust secretion of IL-468 1β that was sensitive to colchicine, and thus dependent on Pyrin (Fig. 5a). By 469 comparison, Nigericin mediated CP-456,773-sensitive IL-1β release after both 470 . CC-BY 4.0 International license made available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprint this version posted July 7, 2021. ; https://doi.org/10.1101/2021.07.07.451422 doi: bioRxiv preprint 3 and 18 hours of LPS priming but was not affected by colchicine (Fig. 5a). 471 Similarly, priming the cells for 18h, but not 3h, with IFN-β was sufficient to 472 trigger Pyrin-dependent IL-18 secretion, demonstrating that it could also prime 473 a Pyrin response (Fig. 5b). To determine whether the increase in Pyrin expression was a requirement for 518 Pyrin activation, we transfected hMDM with two distinct siRNA against Pyrin or 519 a scrambled control 24h before priming with LPS. Notably, both Pyrin targeting 520 . CC-BY 4.0 International license made available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprint this version posted July 7, 2021. ; https://doi.org/10.1101/2021.07.07.451422 doi: bioRxiv preprint siRNAs effectively prevented the LPS mediated increase in Pyrin expression. 521 Still, they did not reduce it further than the untreated control, while the 522 scrambled control had no effect (Fig. 5d). siRNA transfected hMDM were 523 primed with LPS for 18h and then stimulated with TcdA or Nigericin. We 524 observed that the two Pyrin siRNAs, but not the scrambled control, prevented 525 TcdA-mediated IL-1β release. In contrast, neither of the Pyrin-targeting siRNAs, 526 nor the control siRNA, affected Nigericin-mediated inflammasome activation 527 (Fig. 5d), establishing decreasing Pyrin expression is sufficient to inhibit Pyrin 528 activation specifically. 529 530 As Pyrin expression increased after treatement with LPS or interferons, but not 531 Pam3Cys4K or TNFa, we hypothesised that the increase in LPS-dependent 532 Pyrin expression likely required the TRIF signaling pathway. We tested our 533 hypothesis by investigating whether blocking TLR4-mediated TRIF signaling 534 prevents the LPS-dependent increase in Pyrin expression and restores the 535 Pyrin inflammasome response. To this end, we pretreated hMDM with 536 pepinhTRIF, a peptide that prevents the interaction of TRIF with its downstream 537 interaction partners. We then incubated these cells with LPS for 18h before 538 assessing Pyrin expression and activation. We found that treatment with 539 pepinhTRIF, but not a control peptide, reduced LPS mediated Pyrin expression 540 (Fig. 5e). We then stimulated the cells with either TcdA or Nigericin. We found 541 that only the TcdA-mediated inflammasome response was inhibited by 542 pretreatment with the pepinhTRIF, whereas the Nigericin-mediated IL-1β 543 secretion was unaffected (Fig. 5e). These results demonstrate that the LPS-544 . CC-BY 4.0 International license made available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprint this version posted July 7, 2021. ; https://doi.org/10.1101/2021.07.07.451422 doi: bioRxiv preprint stimulated increase in Pyrin expression is TRIF-mediated, and blocking this is 545 sufficient to prevent Pyrin activation in these cells. 546 547 Given that increased Pyrin expression is necessary for Pyrin reactivation in 548 hMDM, we next sought to determine whether Pyrin overexpression alone would 549 enable a Pyrin inflammasome response. For this experiment we used the 550 caspase-4, NLRP3 KO BLaER1 cells, which otherwise do not mount an 551 inflammasome response to TcdB. We overexpressed Pyrin in the caspase-4, 552 NLRP3 KO BLaER1 cells using reconstitution with a vector alone as a control. 553 We then primed these cells with LPS and stimulated them with TcdB, Nigericin 554 or needletox. Markedly, TcdB mediated an inflammasome response in the 555

Pyrin-reconstituted cells, but not in cells transduced with the vector alone 556
Notably, this was not inhibited by CP-456,773 (Fig. 5f). In contrast, Pyrin 557 reconstitution had no effect on either NLRP3 or NLRC4 activation (Fig. 5f). This priming, demonstrating that this is unlikely to be the mechanism preventing 580 Pyrin activation (Fig. 6). The results of our study demonstrate that regulatory mechanisms governing 614 Pyrin activation in hMDM prevent Pyrin activation unless the cell has been 615 exposed to a prior pro-inflammatory stimulus. Notably, in the case of Pyrin, 616 licensing required prolonged stimulation with LPS or interferons, but not other 617 inflammatory stimuli including TNFa. This contrasts with the data from a 618 previous study in PBMCs, where TNFa and Pam3CS4K stimulation also 619 . CC-BY 4.0 International license made available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprint this version posted July 7, 2021. Pyrin associated auto-inflammatory conditions, where, compared to NLRP3 635 driven auto-inflammatory conditions, IL-1β blocking therapy is only somewhat 636 successful in preventing disease [26]. Our data suggests that interferon 637 signaling could also play a role in triggering auto-inflammation in these patients, 638 as well as DAMPs released after tissue injury that stimulate TLR4. Further 639 research is needed to understand the contribution of these sensors to Pyrin 640 based auto-inflammatory conditions to determine if blocking these priming 641 signals helps to control these diseases. 642 643 . CC-BY 4.0 International license made available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprint this version posted July 7, 2021. between these cell types, which also differ in their responses to LPS, which 665 activates NLRP3 in monocytes without a need for a second stimulus, as well as 666 dsDNA, which triggers a STING-and NLRP3-dependent inflammasome 667 response in monocytes rather than activating AIM2 [23,28]. That Pyrin is active 668 . CC-BY 4.0 International license made available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprint this version posted July 7, 2021. ; https://doi.org/10.1101/2021.07.07.451422 doi: bioRxiv preprint in monocytes is unsurprising, as they are migratory cells that rely on actin 669 rearrangement to reach sites of infection where they contribute to the immune 670 response and clearing the pathogen. Therefore, inhibition of migration 671 represents a disruption of a basic function of these cells. This is consistent with 672 Pyrin activation in neutrophils, another migratory immune cell type. Conversely, 673 our results demonstrate that Pyrin is unable to respond in hMDM, even though 674 inactivation of Rho will also impact the immune response and viral clearance. 675 This suggests a requirement for more nuanced control of Pyrin activation in 676 hMDM, which may be a requirement to limit Pyrin driven auto-inflammation. 677 Limiting the Pyrin response to situations requiring pre-exposure to pro-678 inflammatory molecules such as LPS or interferons may be necessary to 679 ensure that Pyrin is not aberrantly activated by endogenous molecules such as 680 bile acids [29], which may be present in tissues but not in the circulation [30]. (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprint this version posted July 7, 2021. indeed the mechanism holding Pyrin inactive requires the less well 706 characterized mechanism involving the B30.2 domain, our data suggest that 707 this control mechanism will be disrupted by these mutations. CC-BY 4.0 International license made available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprint this version posted July 7, 2021. ; https://doi.org/10.1101/2021.07.07.451422 doi: bioRxiv preprint

Inflammasome stimulation assays 793
Primary monocytes/hMDM: Cells were harvested and seeded the day before 794 the assay. Before the experiment, the media was removed, and fresh media 795 with or without a TLR stimulus was added (LPS 10ng/ml, Pam3CS4K 25 ng/ml) 796 and incubated for 3h. Next, compounds were added and incubated for 15 min 797 with 100 ng/mL LPS for 3h, followed by treatment with 8 uM Nigericin (NLRP3) 815 or 100 ng/mL PA + 200 ng/mL LFn-PrgI (needletox, NLRC4) for 1.5 h, or 2 816 µg/mL TcdB for 8h. Supernatants were cleared by centrifugation at 4° C, 1000g 817 . CC-BY 4.0 International license made available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprint this version posted July 7, 2021. Next, the upper phase was discarded, the same volume of methanol of the 835 previous step was added, and the sample was centrifuged for 3 min at 13,000g. 836 The pellet was then dried and resuspended in 1x LDS-sample buffer containing 837 a 10% sample reducing agent (Invitrogen). Samples were heated at 95°C for 5 838 minutes and collected by centrifugation before loading. 839 840 Proteins were separated by 4-12% SDS-PAGE in precast gels (Novex; 841 Invitrogen) with MOPS buffer for proteins above 50kDa or MES buffer for 842 . CC-BY 4.0 International license made available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprint this version posted July 7, 2021. CC-BY 4.0 International license made available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprint this version posted July 7, 2021. The cells were centrifuged at 2500g for 10 min. After centrifugation, the (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprint this version posted July 7, 2021.