The helminth T2 RNase ω1 promotes metabolic homeostasis in an IL-33– and group 2 innate lymphoid cell–dependent mechanism

Induction of a type 2 cellular response in the white adipose tissue leads to weight loss and improves glucose homeostasis in obese animals. Injection of obese mice with recombinant helminth-derived Schistosoma mansoni egg-derived ω1 (ω1), a potent inducer of type 2 activation, improves metabolic status involving a mechanism reliant upon release of the type 2 initiator cytokine IL-33. IL-33 initiates the accumulation of group 2 innate lymphoid cells (ILC2s), eosinophils, and alternatively activated macrophages in the adipose tissue. IL-33 release from cells in the adipose tissue is mediated by the RNase activity of ω1; however, the ability of ω1 to improve metabolic status is reliant upon effective binding of ω1 to CD206. We demonstrate a novel mechanism for RNase-mediated release of IL-33 inducing ILC2-dependent improvements in the metabolic status of obese animals.— Hams, E., Bermingham, R., Wurlod, F. A., Hogan, A. E., O’Shea, D., Preston, R. J., Rodewald, H.-R., McKenzie, A. N. J., Fallon, P. G. The helminth T2 RNase ω1 promotes metabolic homeostasis in an IL-33– and group 2 innate lymphoid cell–dependent mechanism.

Obesity is associated with a low-grade proinflammatory state, with a central role for immune cells that infiltrate the adipose tissue contributing to localized and systemic inflammation and promoting insulin resistance and the development of metabolic syndrome (1,2). The extent of obesity-related metabolic dysfunction directly correlates with the recruitment of proinflammatory cells, M1 classically activated macrophages (CAMs), and CD8 + TandCD4 + Thelper(T h )-1 cells into the adipose tissue. Conversely, in lean individuals, the adipose tissue is populated with eosinophils, regulatory T (T reg ) cells, NK T cells, group 2 innate lymphoid cells (ILC2s), and M2 alternatively activated macrophages (AAMs), which promote insulin sensitivity and metabolic homeostasis (3)(4)(5).
In experiments on obese mice, the artificial generation of a type 2 environment (e.g., by infecting with type 2inducing helminths such as Nippostrongylus brasiliensis and Schistosoma mansoni) has been shown to induce sustained weight loss and improvement in both glucose tolerance and insulin sensitivity (3,4,6). The helminth-mediated modulation was associated with a concomitant increase in eosinophil and ILC2 recruitment within the epididymal white adipose tissue (E-WAT) of the mice and the release of the type 2 cytokines IL-4, IL-5, and IL-13, resulting in increased AAMs (3,4,6). Furthermore, injection of obese mice with type 2-inducing Schistosoma egg-derived antigens also improved metabolic homeostasis (6). Analysis of soluble antigens in S. mansoni eggs identified the glycoprotein S. mansoni egg-derived v1( v1), a T2 RNase (7), as the predominant type 2-inducing component (8,9).
In this study, we sought to address the mechanisms underlying type 2 modulation of obesity and metabolic homeostasis using v1 as a type 2-inducing molecule. Treatment of obese mice with v1 induced a potent type 2 cellular response in the adipose tissue, with associated release of IL-33 from adipocytes, stimulating cells within the Abbreviations: a1/IPSE, Schistosoma mansoni egg-derived a1; v1, Schistosoma mansoni egg-derived v1; AAM, alternatively activated macrophage; CAM, classically activated macrophage; DC, dendritic cell; E-WAT, epididymal white adipose tissue; IL-33R, IL-33 receptor; ILC2, group 2 innate lymphoid cell; OVA, ovalbumin; ROR, retinoic acid orphan receptor; T h , T helper; T reg , regulatory T; UCP1, uncoupling protein 1; WAT, white adipose tissue; WT, wild-type peritoneal cavity and adipose tissue to release type 2 cytokines, resulting in a localized accumulation of innate type 2 cells in the E-WAT. This capacity of v1toinduceIL-33release and modulate obesity is dependent upon the molecule's RNase activity. Data presented herein demonstrate a potential RNase-mediated mechanism for modulation of type 2 immunity by an IL-33-dependent ILC2-mediated mechanism to alter obesity and stabilize glucose homeostasis.

MATERIALS AND METHODS
Animals C57BL/6J, Cd206 2/2 ,a n dRora sg/sg mice were purchased from The Jackson Laboratory (Bar Harbor, ME, USA). T1/ST2deficient mice (Il1rl1 2/2 ) (10), IL-33 citrine reporter mice (Il33 Cit/+ ), and IL-33-deficient mice (Il33 Cit/Cit ) (11) were crossed to C57BL/6J in-house. Conditional retinoic acid orphan receptor (Ror)a floxed mice were generated, and Rora fl/sg were crossed to Il7r Cre (12), as previously described (13). Animals were housed in a specific pathogen-free facility in individually ventilated and filtered cages under positive pressure. All animal experiments were performed in compliance with the Irish Medi cines Board regulations and approved by the Trinity College Dublin's BioResources ethical review board.

High-fat diet and in vivo metabolic testing
Age-matched mice were fed a high-fat diet (60% kcal fat, D12492; Research Diets, Inc., New Brunswick, NJ, USA) or control diet (10% kcal fat) for 8-16 wk as indicated (14). Glucose tolerance was assessed in mice unfed overnight and challenged with 2 g/kg glucose i.p. Insulin tolerance was tested in mice unfed for 4 h and challenged with 0.75 mU/g human insulin i.p. Blood glucose was measured at the times indicated using a glucometer (Abbott Laboratories, Abbott Park, IL, USA).

S. mansoni egg antigen injection
S. mansoni eggs were isolated from the livers of infected mice and soluble egg antigens prepared as previously described (15). Native a1andv1 were isolated from soluble egg antigens by column chromatography, as previously described (16). Obese mice were injected intraperitoneally with 25 mg native a1(d0,2,and4)or 25 mg native v1 (d 0, 2, and 4), and weight was monitored, as indicated in the figure legend (Supplemental Fig. S1).
Corresponding groups of mice were treated with 25 mg endotoxin-free ovalbumin (OVA) as a glycoprotein control.
Histology E-WAT from mice was perfused and fixed in 10% formalin saline, and then paraffin embedded. Paraffin-embedded sections were cut to 4 mM and the slides stained with hematoxylin and eosin. For uncoupling protein 1 (UCP1) staining, slides were deparaffinized, and antigen retrieval was carried out by heating sections to 95°C in sodium citrate buffer. Sections were blocked with goat serum before incubation with rabbit anti-UCP1 antibody (ab10983; Abcam Inc., Cambridge, United Kingdom) diluted 1:100. Following peroxidase blocking, horseradish peroxidase-conjugated goat anti-rabbit (P0448; Dako, Glostrup, Denmark) was used as the secondary antibody, and sections were incubated at 1:1000 in PBS for 1 h at room temperature. For staining, diaminobenzidine chromagen (K3468; Dako) was used according to the manufacturer'sinstructions,and Mayer'shematoxylinwasusedtocounterstain.

Murine cell culture and stimulation
Murine adipocytes and macrophages were isolated from adipose tissue after incubation with 1 mg/ml collagenase D from Clostridium histolyticum (Roche Applied Science, Burgess Hill, United Kingdom). Briefly, the E-WAT was collected, mechanically shredded, and incubated with collagenase D at 37°C with gentle shaking for 1 h. After centrifugation at 400 g, the surface adipocyte fraction was gently collected and washed 3 times in PBS supplemented with 10% fetal calf serum. Adipocytes were identified as SSC hi FSC hi CD90 + Sca-1 + CD11b 2 by flow cytometry. Macrophages (F4/80 + cells) were prepared from the stromal vascular fraction from the E-WAT. Adipocytes and macrophages were cultured in Roswell Park Memorial Institute medium (Buffalo, NY, USA) supplemented with 2 mM L-glutamine, 100 U/ml penicillin, and 100 mg/ml strept o m y c i na tad e n s i t yo f23 10 6 cells/ml for 24 h, and the culture supernatant was discarded. Cells were incubated with 500 ng/ml recombinant v1for1-24 h, as indicated in the figure legend ( Fig.  4E). Culture supernatants were collected for cytokine profiling.

Serum triglyceride quantification
Serum was assessed for triglyceride levels using the Abnova triglyceride quantification kit (Heidelberg, Germany) following the manufacturer's instructions.

Mouse cytokine ELISAs
Cytokine levels were quantified in tissue culture supernatants, peritoneal lavage fluid samples, and serum. Samples were analyzed by sandwich ELISAs to quantify levels of specific cytokines. IL-33 was measured with the DuoSet ELISA development system from R&D Systems (Abingdon, United Kingdom) following the manufacturer's protocol.

Human adipocyte isolation
Adipocytes were isolated from omental adipose tissue biopsies from obese patients (body mass index ,50) undergoing elective bariatric surgery. Clinical studies were approved by the St. Vincent's University Hospital, Dublin Ethics Committee. Written informed consent was obtained from each participant before commencement of research activities. Omental samples were processed to isolate adipocytes as described above. Adipocytes were incubated with 500 ng/ml v1 for 3 and 24 h. IL-33 was quantified in the culture supernatant by ELISA (R&D Systems).
Flow cytometry E-WAT from lean and obese mice was mechanically shredded and incubated with 1 mg/ml collagenase D from C. histolyticum and a single-cell suspension prepared from the stromal vascular fraction (5). In addition, the adipocytes were collected and prepared for flow cytometry. Surface marker expression was assessed by flow cytometry with data collection on a CyAn (Beckman Coulter, High Wycombe, United Kingdom), and data were analyzed using FlowJo software (Treestar, Ashland, OR, USA). IL-33 expression was determined using Il33 Cit/+ reporter mice. To identify ILC2s, cells were stained with BD Biosciences ( Using appropriate isotype controls, quadrants were drawn, and data were plotted on logarithmic scale density plots.

RNA isolation and real-time PCR
RNA was isolated from homogenized E-WAT using the RNeasy kit and reverse transcribed using the QuantiTect Reverse Transcription Kit incorporating a genomic DNA elimination step (Qiagen, Germantown, MD, USA). Real-time quantitative PCR was performed on an ABI Prism 7900HT sequence detection system (Applied Biosystems, Dublin, Ireland) using predesigned TaqMan gene expression assays specific for murine UCP1 (Mm01244861_m1) and normalized to murine glyceraldehyde 3-phosphate dehydrogenase. Fold expression was calculated using the DDC t method of analysis.

Statistics
Statistical analysis was performed using GraphPad InStat (GraphPad Software, La Jolla, CA, USA). Results are presented as means 6 SEM. Differences, indicated as 2-tailed P values, were considered significant when P . 0.05 as assessed by unpaired Student's t test with Welch correction applied as necessary.
To further address the immunologic mechanism underlying the ability of v1 to modulate obesity, we generated recombinant v1 from human embryonic kidney 293 cells (Supplemental Fig. S2A). A dose of 25 mg v1 was administered by i.p. injection based on dose response and optimal delivery route experiments (Supplemental Fig. S2C, D). Treatment of obese mice with recombinant v1g i ve n intraperitoneally on d 0, 2, and 4 (Supplemental Fig. S2B) caused a rapid and significant (P , 0.01-0.05) weight loss relative to mice injected with OVA as a control glycoprotein, with peak weight loss on d 6 postinitial injection (Fig.  1A). In addition, there was a significant (P , 0.01-0.001) decrease in adiposity, characterized by a decrease in the weight of both E-WAT and inguinal WAT in v1-treated obesemice (Fig.1B). This was also reflected in a significant (P , 0.01) reduction in adipocyte area of v1-treated obese mice ( Fig. 1C). Lean mice treated with v1 had no alterations in weight gain, and additionally, food and water intake was comparable between v1-or OVA-treated lean or obese mice (data not shown). Analysis of the effects of v1 on metabolic parameters demonstrated that v1 treatment significantly (P , 0.01) lowered fasting blood glucose in obese mice to a level comparable to mice fed a control diet (Fig. 1D), with significant (P , 0.05) decreases in glucose tolerance (Fig.  1E), although there was no significant improvement in insulin sensitivity (Fig. 1F). Consistent with the improvements in metabolic function, there was a decrease in serum triglyceride levels in obese mice treated with v1 (Fig.1G).
v1i sag l y c o p r o t e i n ,a n di t sa c t i o ni sp a r t l ym e d i a t e d through its binding to the mannose receptor (CD206) on the cell surface (17). CD206 is expressed on both AAMs and adipocytes in the E-WAT ( Fig. 2A)o fo b e s em i c e , suggesting that v1 could act directly on such cells within the adipose tissue. Indeed, when injected into obese mice unable to signal via CD206 [Cd206 2/2 (20)], v1d i dnot cause weight loss (Fig. 2B). Furthermore, using recombinant v1 with mutations in the sites responsible for glycosylation (N71/176Q; v1 DGLY ), we show no effect on weight gain (Fig. 2C). These data indicate that binding of v1t o CD206 is essential for its capacity to induce weight loss and alter the metabolic status of obese mice.
It has become apparent that the cellular composition of the adipose tissue is also partly responsible for the metabolic status of both mice and humans (21). Although WAT is an effective energy store, brown or beige adipocytes are responsible for regulating calorific expenditure, thus preventing weight gain in mice and humans (22,23). The weight loss and metabolic changes in v1-treated mice were associated with an increase in the expression of UCP1, a surrogate marker for beiging of WAT (24,25), in the WAT (Fig. 1I). Therefore, v1 improves metabolic function and inducesbeigecellswithintheadiposetissue.
v1 induces localized type 2 cellular responses in the E-WAT of obese mice Several studies have shown that the induction of a type 2 response, either systemically by infection (3,6) or by injection of an exogenous cytokine (5), can induce alterations in the immune cell populations in the adipose tissue of obese mice leading to improved glucose homeostasis. Because v1 is associated with the induction of IL-4 and IL-5 (8,9,17), we analyzed changes in the immune cell profile oftheE-WATinobesemiceinjectedwithv1atboth6dand 21 d after initial treatment following intraperitoneal injections on d 0, 2, and 4 (d +2 and d +17, respectively). In obese mice injected with the control protein OVA, the cellularity of the E-WAT was proinflammatory, with T h 1c e l l sa n d CAMs predominating (Fig. 3A, B). ; P , 0.001) in the E-WAT of obese mice. These data demonstrate that the weight loss and downstream improvement in glucose metabolism in response to v1are associated with an increase in eosinophils and ILC2s in the WAT and with the induction of AAM polarization; these are cellular changes in adipose tissue known to impact on metabolic status (3,26). Furthermore, these alterationsincellularcompositioncanstillbedetected in the adipose tissue 17 d after treatment has ceased; although these cellular increases are no longer significant, these data do suggest that v1h a sas u s t a i n e d beneficial effect on the cellular milieu in the E-WAT.

v1 induces the systemic and localized release of IL-33
Because v1 induces potent type 2 responses, the role of the epithelial cell-derived type 2-promoting cytokines IL-33 and thymic stromal lymphopoietin was investigated. The ability of IL-25 to drive ILC2s and type 2 NK T cells, res ul tinginwei ghtlossinobes emice,hasbeenreported (5), and IL-33 is also involved in maintaining glucose homeostasis and promoting beiging of WAT (25,27). In mice injected with v1, whereas we do not observe significant induction of IL-25 or thymic stromal lymphopoietin, IL-33 is released into the peritoneal cavity 3 h postinjection ( Fig.  4A; data not shown). IL-33 is a member of the IL-1 superfamily, typically released by stromal cells, mast cells, and dendritic cells (DCs) (28). Using an IL-33 reporter mouse (11), we analyzed the cellular repertoire expressing IL-33 in the E-WAT. Although IL-33 expression was apparent in both macrophages and DCs to a small extent, the primary cellular source of IL-33 in the E-WAT was adipocyte (SSC hi FSC hi CD90 + Sca-1 + CD11b 2 )( F i g .4 B). Analysis of the kinetics of IL-33 expression demonstrates maximal expression of IL-33 in adipocytes 24 h after injection of v1, whereas macrophages increase expression of IL-33 72 h after treatment (Fig. 4C). Furthermore, in vitro stimulation with v1 induced significant (P , 0.0001) IL-33 release from adipocytes isolated from obese mice (Fig. 4D)a n d humans (Fig. 4E). Using CD206-deficient mice (Cd206 2/2 ), we show that CD206 expression on the adipocyte is required for IL-33 release following in vitro treatment with v1 (Fig. 4D). Interestingly, macrophages isolated from the E-WAT from either wild-type (WT) or Cd206 2/2 mice do not release significant levels of IL-33 in response to in vitro culture with v1; although expression of IL-33 is increased in these cells in vivo (Fig. 4C). The release of IL-33 from cells is often associated with cell death. In accordance with the cytotoxic nature of v1, we observe increased death of peritoneal exudate cells in the peritoneal cavity following v1t r e a t m e n t( F i g .4 E). Analysis of the IL-33expressing adipocytes demonstrated an increase in cell death associated with v1treatment (Fig.4F). These data suggest that the localized cytotoxicity of v1i n d u c e s E-WAT adipocytes to increase production and release of IL-33.

RNase activity of v1 induces weight loss and regulates glucose homeostasis
v1hasbee nidentified as a T2 RNase (7) (Supplemental Fig. S2E), a property shown to be integral to the ability of v1 to induce IL-4 and IL-5 release (17). A recombinant v1R N a s e -n u l l( v1D RNase ) mutant was generated, by substituting a phenylalanine residue in the RNase catalytic domain with a histidine residue (H58F), that was devoid of RNase activity (Supplemental Fig. S2E).
Treating obese mice with v1D RNase did not induce significant weight loss or a signific a n tr e d u c t i o ni na d iposity (Fig. 5A, B). Furthermore, v1D RNase did not improve glucose tolerance in obese mice (Fig. 5C). The absence of functional RNase activity also decreased type 2cellinfiltration into the E-WAT, with fewer ILC2s and AAMs observed; although interestingly, eosinophil infiltration was still significantly (P , 0.001) increased (Fig.  5D). In contrast to the mutant v1D RNase protein, recombinant v1 with RNase activity (Supplemental Fig.  S2E) was efficacious in modulating these parameters in obese mice (Fig. 5A-D). Furthermore, RNase activity was required for v1-elicited release of IL-33 into the peritoneal cavity of mice (Fig. 5E) as well as inducing IL-33 expression in adipocytes (Fig. 5F). These data confirm a role for the RNase activity of v1 in the induction of metabolic changes and also demonstrate an RNasemediated mechanism for the stimulation of the release of IL-33.

Weight loss and type 2 cell induction by v1a r e dependent on IL-33
To formally validate a role for IL-33 in v1-mediated weight loss, obese IL-33 receptor (IL-33R; T1/ST2)deficient (Il1rl1 2/2 ) and IL-33-deficient (Il33 Cit/Cit ) mice were injected with v1. In the absence of IL-33 and    IL-33-mediated signaling, v1 did not induce significant weight loss in obese animals as compared to WT animals (Fig. 6A). Furthermore, there was no alteration in glucose tolerance in IL-33 pathway-deficient obese mice treated with v1( F i g .6 B). In support of v1-induced weight loss and the changing cellularity of E-WAT being IL-33 dependent, injection of v1 into obese IL-33-and IL-33R-deficient mice did not significantly induce ILC2s, eosinophils, or AAMs in the E-WAT (Fig. 6C). We next investigated the identity of the IL-33-responsive cells responsible for weight loss in the v1-treated mice. IL-33 is a potent inducer of ILC2s in mice, and these cells have been identified recently as playing important roles in the activation and recruitment of eosinophils and AAMs in the E-WAT (4, 25). ILC2s are reliant upon the transcription factor RORa for their development, and staggerer mice (Rora sg/sg ), which have a natural mutation in the gene encoding RORa, are specifically deficient in ILC2s (29). ToaddresstheroleofILC2sinv1-induced weight loss, ILC2-deficient Rora fl/sg Il7r Cre mice were used (13). These mice demonstrate a complete ablation of ILC2s (Fig. 7A), in the context of this study we do not see significant reductions in the basal levels in any other cell population, however, the depletion of ILC2s can E % of Max.  (Fig. 7A). Furthermore, the ability of v1 to improve glucose tolerance was ablated in mice deficient in ILC2s (Fig. 7A). In addition, in the absence of ILC2s, injection of obese mice with v1fa il edtoin du cein filtration of eosinophils and AAMs into the E-WAT of obese mice (Fig. 7B). These data indicate that v1 requires IL-33 and ILC2s to cause the alteration in the immune cell milieu in the E-WAT, resulting in increased eosinophilia and localized AAM polarization, which are associated with an improvement in metabolic parameters.

DISCUSSION
Obesity and metabolic disorders are rapidly becoming an epidemic and a major cause of morbidity in the developed world. Many studies have focused on the immunologic causes of obesity and have associated the obese state with a low-grade inflammatory response in the adipose tissue (30). By promoting a more type 2 anti-inflammatory environment both systemically and in the adipose tissue, weight loss and an improvement in metabolic status can be achieved (3-5, 24, 25). Data presented herein identify a role for an isolated helminth egg-derived T2 RNase, v1, to improve metabolic homeostasis in obese animals by a mechanism dependent on driving IL-33 release and localized ILC2 recruitment.
Studies have previously used helminths as a method of inducing a systemic type 2 response and have demonstrated favorable effects on metabolic parameters in mice (3,6). This return to metabolic homeostasis in helminthinfected obese mice is associated with increased eosinophils, ILC2s, and AAMs in the adipose tissue, a change in immune status that can be maintained through chronic infection (3,6). This study provides mechanistic insight into helminth modulation of obesity and identifies a specific protein released by helminth eggs, which is capable of driving a similar alteration in immune and metabolic parameters as that observed in live helminth infection. Studies have focused on the ability of v1 to drive DCs to polarize naive CD4 cells to a T h 2 phenotype, via a mechanism requiring both glycosylation and RNase activity (17). Although glycosylation is required for internalization of v1 via binding to CD206, the RNase interferes with protein synthesis by global cleavage of rRNA and mRNA once translocated to the cytosol, enabling v1 to condition DCs for priming of T h 2 responses (17). The v1 RNase is not unique in its association with induction of a T h 2 response; the birch  pollen antigen Bet v-1 has been identified as an RNase (31), and the fungal RNase Aspf-1 has allergenic capacity (32). These studies suggest that the ability to induce a T h 2 response via RNase activity may not be unique to helminths.
The mechanisms described in this study identify a central role for IL-33 in obesity and metabolic homeostasis. Indeed, IL-33 is rapidly becoming accepted as an important factor in limiting obesity and metabolic dysregulation with several groups identifying a central role for IL-33 in metabolism (25,27,33). We have shown that v1isable to drive IL-33 release from cells, in particular adipocytes, through its intrinsic RNase activity. This v1-mediated release of IL-33 not only alters the immune environment within the adipose tissue but also promotes the beiging of WAT, in a mechanism that limits adiposity by increasing calorific expenditure (24). Furthermore, we have shown that, in addition to immune cells within the WAT, adipocytes themselves are a target for v1 and express and release IL-33. This appears to be mediated via the RNase function of the protein because IL-33 expression is ablated if the RNase function of v1 is mutated. However, v1isahepatotoxin (18,34), and IL-33 release is often associated with cellular necrosis (35). Indeed, we do observe localized cell death in vitro in response to v1 treatment, although at low levels. Furthermore, ,50% of the IL-33-expressing 0246 Time (days) 107.5 102.5 97.5 Weight gain (%) 92  adipocytes are dying; therefore, whereas cell death is likely to be a mechanism inducing IL-33 and thus causing weight loss, there is potentially another mechanism underlying IL-33 release in response to v1.
IL-33 is an important inducer of ILC2, a cell widely implicated in improving metabolic status in obese mice (4,5,25). Notably, we also observed that v1-induced ILC2s were required for weight loss and improving glucose tolerance. However, due to the lack of CD206 on the surface of ILC2s (36), it is unlikely that v1 acts directly on ILC2s. Instead, we propose a mechanism where v1b i n d i n gt oC D 2 0 6o n adipocytes initially induces IL-33 release in part via an RNase-mediated mechanism, thereby inducing ILC2s and other innate cells to release IL-4, IL-5, and IL-13 resulting in a switch in polarization of macrophages to AAMs (Fig. 8).
This study provides further mechanistic insight into the immune-mediated regulation of obesity. We demonstrate an important role for RNase-mediated IL-33 release in promoting metabolic homeostasis in obese animals. We also identify the ability of v1 to induce IL-33 release from human adipocytes, suggesting that any beneficial effect that v1 treatment has in mice has the additional potential to be beneficial in humans. Data presented herein not only provide novel mechanistic insights into the roles of IL-33 and ILC2s in promoting metabolic homeostasis, but they also raise the potential beneficial therapeutic use of type 2-inducing RNases in the treatment and management of obesity and obesityrelated metabolic disorders. This work was supported by Science Foundation Ireland and the National Children's Research Centre.