A helminth chitinase structurally similar to mammalian chitinase displays immunomodulatory properties

Previously, we reported significant immunomodulatory effects of the entire excretory-secretory (ES) proteins of the first larval stage (L1) of the gastrointestinal nematode Trichuris suis in a rodent model of allergic hyperreactivity. In the present study, we aimed to identify the proteins accounting for the modulatory effects of the T. suis L1 ES proteins and thus studied selected components for their immunomodulatory efficacy in an OVA-induced allergic airway disease model. In particular, an enzymatically active T. suis chitinase mediated amelioration of airway hyperreactivity, primarily associated with suppression of eosinophil recruitment into the lung. The three-dimensional structure of the T. suis chitinase as determined by high-resolution X-ray crystallography revealed significant similarities to mouse acidic mammalian chitinase (AMCase). In addition, the unique ability of T. suis chitinase to form dimers, as well as acidic surface patches within the dimerization region may contribute to the formation of cross-reactive antibodies to the mouse homologs. This hypothesis is supported by the observation that T. suis chitinase treatment induced cross-reactive antibodies to mouse AMCase and chitinase-like protein BRP-39 in the AHR model. In conclusion, a biologically active T. suis chitinase exhibits immunomodulatory properties despite its structural similarity to the mammalian counterpart. Author summary Experimental immunotherapy via reintroduction of intestinal worms to treat and prevent autoimmune, chronic inflammatory or allergic diseases is being discussed but the underlying mechanisms are still not fully understood. Here, we investigated the immunomodulatory potential of specific proteins of the whipworm Trichuris suis that are secreted very early during larval development. Using a murine model of allergic lung disease, we show that in particular one T. suis protein, functionally characterized as an active chitinase, is reducing the lung inflammation. The T. suis chitinases three-dimensional protein structure revealed remarkable similarities to the hosts’ chitinase, an enzyme known to play a pivotal role in lung allergy. We also show that treatment with the helminth chitinase induced cross-reactive antibody responses against murine chitinase and chitinase-like proteins, both being inflammatory marker and regulators of type 2 immunity. Thus, our study provides a novel mechanism of immunomodulation by helminth components and may contribute to a better understanding of clinical responses of patients receiving helminthic therapy.


Nonstandard abbreviations
The genes corresponding to selected T. suis proteins were amplified and cloned into the vector 265 Helmholtz-Zentrum Berlin (38) and processed using XDSAPP (39). Relevant processing 327 statistics are shown in Table 2. The structure was solved by molecular replacement using the 328 structure of Human Chitotriosidase (CHIT1) as a search model (PDB-Id 5HBF, (40)). After 329 several cycles of manual rebuilding using COOT (41) and refinement using REFMAC5 (42), 330 refinement converged at R-factors of 19 and 23 % for the working R and the free R, respectively 331 (Table 3). The refined structure features good refinement statistics and excellent geometric 332 parameters. The refined coordinates and the associated structure factor amplitudes were 333 deposited in the PDB using the accession code 6G9C. 334 335 Structure based sequence alignment and surface electrostatics. 336 To obtain an overview of structural similarities and differences between Ts-Chit and host 337 chitinases, the topology and secondary structure elements of Ts-Chit were compared to six 338 mouse chitinases and chitinase-like proteins by means of a structure-based amino acid sequence 339 alignment (see Supplementary Information S2). Secondary structure elements were predicted 340 with PSIPRED (43) and a structure-based amino acid sequence alignment was constructed 341 using SBAL (44). 342 In order to assess whether Ts-Chit possesses any features that might serve as a potential 343 molecular mimicry of host chitinases when acting as an antigen, we undertook a comparison of 344 surface features of Ts-Chit with mouse chitinases. In general, this approach is limited by the 345 availability of only one experimental 3D structure of mouse chitinases. Using the amino acid 346 sequences of three mouse chitinases (M. musculus AMCase, Chit1 isoform 1 and Ym1), known 347 3D structures or close structural homologues of these proteins were identified in the PDB using 348 pGenThreader (45), revealing mouse AMCase as the only mouse protein suitable for 349 comparison in this context. We previously showed that T. suis excretory/secretory proteins (ES proteins) collected from L1 365 hatched in vitro reduced clinical signs of murine allergic airway disease (33). To identify the 366 early released immunomodulatory proteins responsible for this effect, we now investigated the 367 protein composition of L1 ES proteins and compared it to L2, L3 and L4 larval stages of T. suis 368 released products. For stage-specific expression patterns, parasites were isolated from pigs 369 infected with 8000 TSO at 10, 18 and 28 days post-infection (resulting in L2, L3, and L4 larvae, 370 respectively) and cultivated to collect conditioned media for mass spectrometry. Alternatively, 371 eggs were hatched in vitro to generate L1 larvae and conditioned media, as previously described 372 (33). Comparative proteomic analysis of T. suis ES proteins ( Fig. 1A and Supplementary 373 Table S1) revealed some overlap between first stage (in vitro hatched) L1 larvae and L2 larvae 374 (10 day larvae), but no overlap of L1 ES proteins with later stages of larval development (L3 375 ES -18 day larvae and L4 ES -28 day larvae). Four L1-specific proteins and two L1/L2-specific 376 proteins with predicted signal sequences and a lack of transmembrane domains ( Table 2) were 377 recombinantly expressed. Since glycosylation can be crucial for the immunomodulatory effects 378 of secreted helminth proteins, we used the eukaryotic expression system LEXSY, based on the 379 protozoan L. tarentolae (Fig. 1B). Notably, selected T. suis L1 proteins were distinct in size 380 (ranging from 13 to 56 kDa, Fig. 1C), predicted glycosylation sites and homology-based 381 annotation ( Table 2). 382 We initially screened for immunomodulatory effects of the six recombinant T. suis L1 proteins 383 using the OVA (ovalbumin)-induced allergic airway hyperreactivity model in female BALB/c 384 mice. Mice were treated with recombinant T. suis proteins during sensitization with the allergen 385 OVA on days 0, 7 and 14 (Fig. 1D). The numbers of total BAL infiltrates and BAL eosinophils 386 were detected 2 days after the 2 nd intranasal OVA challenge at day 31 ( Fig. 1D) and analyzed 387 in parallel to control mice lacking OVA sensitization, but receiving OVA-challenge (PBS) and 388 untreated allergic mice (OVA). Analysis of total cell numbers and cellular composition of the 389 BAL fluid found that KFD48490.1 (homologous to T. trichiura acidic mammalian chitinase) 390 and KFD45500.1 (homologous to T. trichiura venom allergen 5) reduced the total numbers of 391 cellular BAL fluid infiltrates, but only KFD48490.1 treated mice showed a significant reduction 392 of eosinophils infiltrating the lungs in comparison to untreated allergic (OVA) mice ( Fig. 1 Fig. 2A), typically observed for active chitinases, and a C-terminal chitin-binding 406 domain with cysteine residues responsible for the chitin binding ( Fig. 2A). Alignment studies 407 additionally proved high sequence similarities to other known helminth chitinases including a 408 putative endochitinase from the pig infecting Ascaris suum and a glycosyl hydrolase from the 409 human hookworm N. americanus (Fig. 2B). Active chitinases cleave chitin from the end of a 410 polymer chain (exochitinase) and/or hydrolyse small oligomers to generate N-411 acetylglucosamine monomers within the oligomer chain (endochitinase) (48). To examine 412 whether KFD48490.1 exerts true chitinase activity, we used two different substrates, 4-MU-413 (GlcNAc)3 and 4-MU-(GlcNAc)2, allowing us to evaluate both endo-and exochitinase activity, 414 respectively (Fig. 2C). KFD48490.1 clearly presented both, endo-and exochitinase activity, 415 whereas extensive, heat-induced proteolysis resulted in a loss of chitinase activity (Fig. 2C). 416 With regard to the biology of nematodes, we explored the influence of low pH and high 417 temperatures hypothesizing that T. suis chitinase will be active at low pH and elevated 418 temperatures in the pig's intestine to support larval hatching and development (49,50). Differential cell counts of BAL fluid further revealed that neutrophil numbers were largely 433 unaffected (Fig. 3C). A trend towards an increase of alveolar macrophages cell numbers was 434 observed in Ts-Chit compared to allergic (OVA) mice (Fig. 3D). Flow cytometry of BAL cells 435 confirmed the reduced frequency of eosinophils (CD45 + CD11b + GR1 low CD11c -SiglecF + ) in Ts-

20
Chit treated compared to untreated (OVA) allergic animals ( Fig. 3E and F), and also supported 437 increased frequencies of alveolar macrophages, identified as CD11b low CD11c + cells among 438 CD45 + cells (Fig. 3G), as observed as a trend in differential cell counts. The respiratory 439 responses resulting from Ts-Chit treatment were measured using non-invasive whole-body 440 plethysmography 24 h after OVA challenge. Importantly, also the non-sensitized control group 441 (PBS) was intranasally challenged with OVA. Preventative administration of Ts-Chit caused 442 restoration of expiration time to the level of PBS control mice at baseline (Fig. 3H). A trend in 443 recovery was also noticed for treating with heat-inactivated Ts-Chit (Fig. 3H). Measuring 444 airway resistance to increasing doses of methacholine (MCh) revealed that Ts-Chit treatment 445 attenuated exacerbation of MCh-induced airway hyperreactivity), but treatment with 446 inactivated Ts-Chit also had some effects (Fig. 3H, I and J). We next evaluated, whether the 447 reduced cellularity and improved lung function observed in Ts-Chit treated animals was 448 reflected in BAL cytokine levels. OVA sensitization and challenge caused a significant increase 449 of the cytokines IL-4, IL-5 and IL-13 detected in BAL fluid; however, Ts-Chit treatment did 450 not reduce but rather enhanced local Th2 cytokine production (Fig. 3K). Interestingly, Ts-Chit 451 and heat-inactivated Ts-Chit alike accelerated IL-18 levels in BAL fluid (Fig. 3K), a cytokine 452 known to induce IL-4, IL-5 and IL-13 release by mast cells and basophils (51). Thus, cellular 453 infiltration and the respiratory response were improved in mice treated with Ts-Chit, but the 454 Th2 cytokine response was unchanged or elevated in comparison to untreated allergic mice 455 To further delineate the direct effects of Ts-Chit treatment on lung inflammation of OVA-457 sensitized and challenged mice, we investigated lung tissue pathology and recruitment of 458 alternatively activated macrophages (AAM). While histopathological findings remained overall 459 similar in the lungs of untreated (OVA) and Ts-Chit treated allergic animals (Fig. 4A), 460 immunohistochemistry revealed increased numbers of interstitial RELMα + cells, a signature 461 marker for AAM that mediates lung vascularization and tissue repair (52,53) when mice were 462 treated with Ts-Chit compared to OVA ( Fig. 4B and C). A similar trend was observed for 463 inactivated Ts-Chit treated animals. Since this was in contrast to our previous observation on 464 Ts-ES L1-induced suppression of RELMα + cells (33) and given that the resistin-like molecule 465 (RELM) a is not only expressed in macrophages but also epithelial cells and tissue infiltrating 466 eosinophils, we used smaller groups of mice to define Relma expression to a interstitial 467 macrophage phenotype by flow cytometry. Lung tissue macrophages were analyzed in 468 homogenates of lavaged lungs and defined as EpCAM -CD11c + CD64 + MHCII + SiglecF - (Fig.  469   4D). While overall frequencies of tissue macrophages were not affected by Ts-Chit treatment 470 ( Fig. 4E), we again detected an increase in Relma + macrophages, but did not reach statistical 471 significance (Fig. 4F). Contrary to the increase of Relma + cells, RT-PCR of homogenized lung 472 tissue of all groups of mice revealed downregulation of arginase-1 and MCP-1 mRNA when 473 mice were treated with intact and also inactivated Ts-Chit compared to PBS (Fig. 4G). 474 Supporting reduced numbers of eosinophil infiltration upon Ts-Chit treatment, we detected 475 downregulation of CCL11/eotaxin mRNA, a potent eosinophil chemoattractant that stimulates 476 recruitment of eosinophils from the blood to sites of allergic inflammation (Fig. 4E). Together, 477 these data illustrate that protein integrity of the Ts-Chit is only partially responsible for the 478 treatment effects leading to attenuated airway hyperreactivity.  Fig. 2). 497 To better understand structural differences between nematode and mammalian chitinases and 498 the structural basis of the serological response to Ts-Chit treatment, we determined the crystal 499 structure of the nematode chitinase protein by X-ray crystallography. The final model analysis 500 data collection and refinement statistics are summarized in Table 3. They indicate that the 501 structure is of high resolution, that it is well refined to very good statistics and hence that it is 502 of high quality. The overall structure reveals a TIM-barrel or (ßa)8-barrel motif, which means 503 that 8 ß-strands form the central barrel which is surrounded by 8 a-helices ( Fig. 5A and C). 504 The active site with the sequence motif D149xD151xE153 is depicted in Figure 5A.2. The crystal 505 structure furthermore revealed a P2-symmetric dimer of the protein, which is stabilized by an 506 intermolecular disulfide bond formed by a cysteine residue at position 180 (Fig. 5A.1 and B). 507 Such dimerization has not been described so far for human or murine chitinases, nor has the 508 occurrence of a cysteine residue at position 180. In addition, the structural alignment with 509 human AMCase (3FXY), human chitotriosidase (1GUV), human YKL-40 (1NWR) and murine 510 Ym1 (1VF8) (Fig. 5C) indicates highly similar structural features in the core regions (marked 511 in red) and certain areas, mostly at the protein surface, where the structure differs between 512 nematode and mammalian chitinases (highlighted in green/blue). A detailed structure based 513 sequence alignment of T. suis chitinase, mouse chitinases and CLPs is given in Supplementary 514

S2.
Notably, the C-terminal region of Ts-Chit from amino acid 401 to 495 contains many 515 threonine repeats and is referred to as unordered region. The T. suis chitinase dimerization was 516 further examined by comparing reducing (+ ß-mercaptoethanol) and non-reducing (-ß-517 mercaptoethanol) conditions for SDS/PAGE loading. We indeed detected a second, higher band 518 at around 75 kDa indicative for dimerization under non-reducing conditions (Fig. 5D). 519 The structural und functional similarity of T. suis chitinase and murine AMCase could either 520 mean that Ts-Chit is sensed similar to murine AMCase resulting in cumulative effects, or in 521 turn, that Ts-Chit interferes with murine AMCase function. Following the idea that T. suis 522 interferes with murine chitinase activity during allergic airway hyperreactivity, we more 523 specifically compared surface features of Ts-Chit and mouse AMCase (PDB: 3FY1), the only 524 acidic mouse chitinase with known experimental 3D structure. To that effect, the surface 525 electrostatics for mammalian AMCase and the Ts-Chit monomer were visualized and appraised. 526 While investigating the monomer alone did not reveal any strikingly identical surface patches, 527 looking at the dimerization region around α3/α4-α3'/α4' (α3:~127-142; α4:~161-180) revealed 528 similarities with an acidic surface patch of AMCase (Fig. 6A). 529 In summary, this first crystallographic structure analysis of a parasitic nematode chitinase 530

T. suis chitinase interferes with murine chitinase in AHR 540
Considering the structural similarities, we asked whether Ts-Chit treatment of allergic mice 541 interfered with the effector functions of its murine counterpart, AMCase. And indeed, chitinase 542 activity was found to be reduced in the BAL fluid of allergic mice treated with intact and 543 inactivated Ts-Chit compared to untreated (OVA) mice (Fig. 7A). Similarly, we detected 544 decreased AMCase transcript levels (Chia1) mRNA transcripts in lung tissues of allergic Ts-545 Chit treated compared untreated (OVA) mice (Fig. 7B). presence of antibodies recognizing Ts-Chit but also murine AMCase and BRP-39 proteins (Fig.  556   7C). In turn, the serum of untreated, allergic mice recognized OVA protein only, but no 557 chitinases or CLPs (Fig. 7C). Together, these data indicate a functional role for cross-reactive 558 antibodies to murine AMCase and BRP-39 that are induced by Ts-Chit treatment. In the context of allergic lung disease, however, observing an active chitinase to mediate 613 suppression of AHR was rather surprising. In recent years, studies in patients and animal models 614 have shown that mammalian chitinases play a key role in mediating the Th2 driven 615 inflammatory responses commonly associated with asthma and have been investigated to 616 indicate disease severity (63,82,83). Mice express two enzymatically active chitinases, acidic 617 mammalian chitinase (AMCase) and chitotriosidase (Chit) and a set of enzymatically inactive 618 chitinase-like proteins (Ym1, Ym2, BRP39). AMCase is secreted by macrophages and 619 epithelial cells of lung and gut. In the lungs, it is constitutively expressed and secreted into the 620 airway lumen. STAT6-activating signals, such as IL-13, further induce AMCase expression and 621 elevated AMCase mRNA and protein levels are detected in BAL of OVA-induced asthmatic 622 mice (63). This latter study also showed that administration of anti-AMCase sera reduced BAL 623 and tissue eosinophilia and BAL chitinase activity. Similarly, in a gut inflammation model it 624 has been shown that the pan-chitinase inhibitor caffeine was able to dampen inflammation in a 625

DSS-induced colitis model (84). 626
Here, we demonstrated that treating mice with a chitinase of T. suis (Ts-Chit) during 627 sensitization reduced BAL eosinophilia, increased numbers of Relma + interstitial cells, 628 improved lung function and reduced chitinase activity in BAL. Interestingly, heat-inactivated 629 Ts-Chit was able to partly phenocopy these effects in vivo. We have no evidence for active Ts- The X-ray crystal structure of Ts-Chit reported here confirmed the expected high similarity of 661 the nematode chitinase with murine acidic mammalian chitinase. In addition, analysis of surface 662 electrostatics of the Ts-Chit dimer suggests the existence of an acidic surface patch on the 663 nematode protein that bears similarity to a surface feature on AMCase monomers. This suggests 664 that the response elicited by Ts-Chit may be due to molecular mimicry; however, future studies 665 need to establish whether this is indeed the case. Site-directed mutagenesis within the 666 dimerization region or the acidic surface patch will be needed to address the impact those 667 structural features to the formation of cross-reactive antibodies.
Molecular mimicry is a phenomenon that is triggered by a high degree of similarity of pathogen 669 and host epitopes and has previously been described for helminth defense molecules of Fasciola 670 hepatica that mimic the molecular function of host antimicrobial molecules (87) or the TGF-ß 671 mimic secreted by Heligmosomoides polygyrus with profound Treg-inducing capacities (19). 672 To the best of our knowledge, a pathogen-driven molecular mimicry inducing antibodies that 673 interfere with Th2-dependent host chitinase activity has not been described so far.