Macrophage-Regulatory T Cell Interactions Promote Type 2 Immune Homeostasis Through Resistin-Like Molecule α

RELMα is a small, secreted protein expressed by type 2 cytokine-activated “M2” macrophages in helminth infection and allergy. At steady state and in response to type 2 cytokines, RELMα is highly expressed by peritoneal macrophages, however, its function in the serosal cavity is unclear. In this study, we generated RELMα TdTomato (Td) reporter/knockout (RαTd) mice and investigated RELMα function in IL-4 complex (IL-4c)-induced peritoneal inflammation. We first validated the RELMαTd/Td transgenic mice and showed that IL-4c injection led to the significant expansion of large peritoneal macrophages that expressed Td but not RELMα protein, while RELMα+/+ mice expressed RELMα and not Td. Functionally, RELMαTd/Td mice had increased IL-4 induced peritoneal macrophage responses and splenomegaly compared to RELMα+/+ mice. Gene expression analysis indicated that RELMαTd/Td peritoneal macrophages were more proliferative and activated than RELMα+/+ macrophages, with increased genes associated with T cell responses, growth factor and cytokine signaling, but decreased genes associated with differentiation and maintenance of myeloid cells. We tested the hypothesis that RαTd/Td macrophages drive aberrant T cell activation using peritoneal macrophage and T cell co-culture. There were no differences in CD4+ T cell effector responses when co-cultured with RELMα+/+ or RELMαTd/Td macrophages, however, RELMαTd/Td macrophages were impaired in their ability to sustain proliferation of FoxP3+ regulatory T cells (Treg). Supportive of the in vitro results, immunofluorescent staining of the spleens revealed significantly decreased FoxP3+ cells in the RELMαTd/Td spleens compared to RELMα+/+ spleens. Taken together, these studies identify a new RELMα regulatory pathway whereby RELMα-expressing macrophages directly sustain Treg proliferation to limit type 2 inflammatory responses.


INTRODUCTION
Macrophages are a dominant resident population within the peritoneal cavity with critical immune surveillance and homeostatic functions (1). As sentinels, they are rapid responders to microbial invasion resulting from injury of the serous organs, such as the spleen, liver and intestinal tract, and can be mobilized to traffic to the injured organ and mediate tissue repair (2). Peritoneal macrophages also perform homeostatic functions to support innate B1 cells (3), clear debris and apoptotic cells (4,5), and dampen inflammation (6)(7)(8). On the other hand, dysregulated peritoneal macrophage responses are associated with diseases including peritonitis, bacterial dissemination, and cancer metastases (9)(10)(11)(12). Identification of peritoneal macrophagederived factors and activation markers that cause beneficial or pathologic outcomes would provide insight into their biology and identify targets for treatment of serous cavity-associated disease.
Peritoneal macrophages, especially the monocyte-derived small peritoneal macrophages, express Resistin-like molecule a (RELMa) under homeostatic conditions (13). In type 2 cytokinepolarized environments such as helminth infection or in vivo IL-4 complex injection, RELMa expression is dramatically elevated reaching 100% expression by small and large peritoneal macrophages (14). RELMa, also known as FIZZ1 and HIMF, was originally identified as a highly secreted protein in the lung during allergic airway inflammation (15), however, it is now well-recognized that RELMa is pleiotropically expressed throughout the body, and a signature gene expressed by M2polarized macrophages in response to multiple helminth infections (16,17). RELMa expression is also triggered by other signals in addition to type 2 cytokines, for example by phagocytosis of apoptotic cells through scavenger receptors (18), or in response to hypoxia (19). Studies in pulmonary inflammation, hypertension and fibrosis, point to an inflammatory function for RELMa by promoting immune cell recruitment, fibroblast activation and proliferation associated with pathogenic fibrosis (20,21). On the other hand, in response to tissue migratory helminth parasites, RELMa critically prevents fatal lung tissue damage, granulomatous inflammation, and promotes tissue repair (22)(23)(24)(25)(26)(27)(28). Downstream regulatory mechanisms for RELMa include limiting CD4 + T cell polarization, promoting anti-inflammatory responses, and mediating collagen cross-linking associated with tissue healing (26,29,30). RELMa also exhibits antibacterial properties by disrupting bacterial membranes for certain bacterial species (31).
Despite high expression levels of RELMa by peritoneal macrophages, whether RELMa affects the role of these cells in immune surveillance or homeostasis is unknown. In this study, we generated transgenic mice where RELMa is deleted and replaced with the TdTomato reporter protein (Ra Td ) and investigated the consequence of RELMa deletion in a polarized type 2 cytokine environment caused by injection of IL-4 complexes. We first validated the Ra transgenic mice and demonstrated successful deletion of RELMa and expression of TdTomato protein, which had an equivalent expression pattern to RELMa. Next, we compared PBS and IL-4c injected Ra +/+ and Ra Td/Td mice, where we identify a critical role for RELMa in limiting IL-4-induced peritoneal macrophage expansion, M1 macrophage activation, and splenomegaly. Gene expression analysis of sorted macrophages from Ra +/+ and Ra Td/Td mice revealed that RELMa deficiency leads the induction of genes promoting T cell response, growth factor and cytokine signaling, but decreased genes associated with differentiation and maintenance of myeloid cells. Combining macrophage-T cell co-cultures, and investigation of ex vivo T cell responses, we further identify a role for macrophage-derived RELMa in promoting regulatory T cell proliferation and the production of IL-10 and GM-CSF. Together, results from these studies validate the utility of Ra Td/Td mice to track RELMa expression and identify a dual role for RELMa in limiting type 2 cytokine immunopathology by cell-intrinsic effects on macrophages and regulatory T cells.

Mice
Retnla Td transgenic mice were generated by genOway (Lyon, France) by homologous recombination in C57BL/6 embryonic stem cells. Retnla Exon 2-4 was targeted using cre recombinase and Flp-mediated excision and replacement with the Td reporter gene, with a WPRE site (32) to enhance reporter expression and stability. The mice were crossed with the genOway proprietary cre-deleter mouse (pCMV driven cre) to generate constitutive Retnla Td mice. Retnla Td/+ heterozygote mice were crossed with C57BL/6 mice to generate littermate homozygote (Td/Td) and WT (+/+) mice after three generations, then bred in-house. Arginase YFP (Yarg) mice were available from Jackson labs. Mice were age matched (6 to 14 weeks old), sex-matched for experiments, and housed under an ambient temperature with a 12 hours light/12 hours dark cycle.

Flow Cytometry and t-SNE Analysis
Peritoneal cavity cells (PECs) were recovered in a total of 5 mL of ice-cold PBS. Splenic macrophage isolation were performed according to previous studies (33).Visceral fat was dissected and single cell dissociation and staining performed as previously described (34). For flow cytometry, cells were blocked with 0.6 µg Rat IgG and 0.6 µg a-CD16/32 (2.4G2) 5 min, stained for 25 min with antibodies to CD11b (M1/70), MHCII (M5/114.15.2), CD11c (N418), CD4(RM4-5), Ly6C(HK 1.4), Ly6G(1A8), CD19(1D3) and CD8(53-6.7) (all from BioLegend, San Diego, CA); SiglecF (E50-2440) (BD Bioscience, San Jose, CA); F4/80 (BM8) (eBioscience, Santa Clara, CA). Cells were analyzed on a Novocyte (ACEA Biosciences, San Diego, CA) or LSRII instrument (BD Bioscience, San Jose, CA) followed by data analysis using FlowJo v10 (Tree Star Inc.; Ashland, OR). t-SNE analyses were performed with FlowJo v10, involving concatenation of samples (5000 cells per biological replicate) from all groups before running the t-SNE analyses to generate plots consistent between groups. This was followed by analysis of separated groups for expression of the desired markers. The parameters used to run the t-SNE analyses are in Supplementary Table 1. Arg, Ra or TdTomato were excluded as parameters given that their expression was being analyzed, and they were negative in the some of the transgenic mouse groups.

Splenocyte Stimulation
Spleens were harvested from PBS or IL4c treated mice at day 4. Single cell suspensions were generated from whole spleen, and red blood cells lysed with ACK lysis buffer. Cells were stimulated in 48 well plates at 5x10 6 cells/well with 1mg/ml of a-CD3 and a-CD28 (eBioscience) as described previously (29). Supernatants were recovered at day 3 for cytokine measurement.

Macrophage and Splenocyte Co-Cultures
Peritoneal cells from naïve Ra +/+ or Ra Td/Td mice were recovered and treated in vitro with IL-4 (20ng/ml) or equivalent control PBS in complete DMEM media (Invitrogen, Gaithersburg, MD). After 24hrs, supernatants were recovered for RELMa ELISA, cells were washed with PBS to remove non-adherent cells, followed by recovery of adherent macrophages with TrypLE ™ Express (Invitrogen, Gaithersburg, MD) and plated in a 96-well flat bottom plate at 2x10 4 cells/well. In vivo-derived M2 macrophages were generated by one i.p. injection of IL-4c, recovery of the peritoneal cells, and F4/80 bead purification using MS columns with >90% purity (Miltenyi Biotec, Inc). Splenocytes were recovered from naïve Ra +/+ mice, and single cell suspensions prepared as above. Splenocytes were CFSE-labelled (5mM, 15 minutes) as previously described (29) (Invitrogen, Gaithersburg, MD), then added to the macrophages (Mac : Splenocyte 1:10) with 0.5mg/ml a-CD3 (5 replicate wells per condition). After 3 and 6 days, nonadherent splenocytes were recovered for flow cytometry analysis on the LSRII (BD Bioscience), and supernatants were recovered for cytokine measurement.

Nanostring Gene Expression Analysis
Peritoneal macrophages (CD11b + F4/80 + ) were sorted with the MoFlo Astrios cell sorter (Beckman Coulter). 5000 macrophages from PBS mice or IL-4c-injected mice were lysed with 1/3 RLT buffer diluted with ddH 2 0 (Qiagen). Lysed cells were processed and quantified by the Myeloid Innate Immunity v2 panel according to manufacturer's instructions (Nanostring). Gene expression analysis was conducted using the Advanced Analysis Nanostring software. Raw counts were normalized to internal controls (4 housekeeping genes, Eef1g, Gusb, Oaz1 and Rpl19), then normalized transcripts with n>30 counts were included for analysis (a total of 309 out of 734 genes). The Nanostring Advanced Analysis algorithm generated biological pathway scores by extracting pathway-level information from a group of genes using the first principal component (PC) of their expression data (35). Pathway scores of Ra +/+ or Ra Td/Td naive and IL-4c-injected mice were analyzed by an unpaired t-test and chosen pathways (p value ≤ 0.05) are represented as the difference in pathway score between the Ra +/+ or Ra Td/Td groups (n=4/group). Differentially expressed (DE) genes (p ≤ 0.05) in each pathway were graphed as heatmaps (36).

Statistical Analysis
Data are presented as mean ± SEM and statistical analysis was performed by Graphpad Prism 9 software. Data was assessed by one-way ANOVA followed by post-hoc Tukey's test for multiple comparison, or by unpaired t-test for 2-group comparisons. For data collected over several time points, two-way ANOVA with post-Sidak multiple test was performed. *, p ≤ 0.05; **, p ≤ 0.01; ***, p ≤ 0.001. Experiments were repeated 2-4 times with n=2-8 per group for in vivo experiments, or 3-5 replicate wells for in vitro studies, apart from Nanostring gene expression analysis, which was performed once (n=2 for naïve and n=4 for IL-4c injected per group).

Generation and Validation of RELMa Transgenic Mice
RELMa is a pleiotropic protein expressed by both immune and non-immune cells, and is detectable in the serum of naïve mice (16). In the serosal cavities, resident macrophages express RELMa in homeostatic conditions, however, RELMa expression is dramatically elevated in a type 2 cytokine environment such as helminth infection or IL-4c injection (17). We sought to address RELMa function in the peritoneal cavity by generating transgenic mice in which cre and flp recombinase mediates RELMa (exons 2-4) deletion and replacement with the TdTomato reporter protein ( Figure 1A). To validate the targeting strategy and enable tracking of RELMa-expressing cells, these founder mice were crossed to a universal cre deleter mouse line so that RELMa expression can be tracked by Td reporter protein in heterozygote mice (Ra Td/+ ), while homozygote mice (Ra Td/Td ) are used to investigate RELMa function. Quantification of RELMa protein in the serum and peritoneal cavity fluid indicated high levels of circulating RELMa under homeostatic conditions in Ra +/+ mice, detectable but significantly reduced RELMa in heterozygote (Ra Td/+ ) mice, and no RELMa in homozygote mice (Ra Td/Td ) ( Figure 1B). RELMa and Td mRNA levels were also quantified in adherent peritoneal cells from naïve mice treated in vitro with IL-4 ( Figure 1C). Both Ra +/+ and Ra Td/+ macrophages had equivalent IL-4 induced RELMa expression. In IL-4 treated Ra Td/Td and Ra Td/+ macrophages, Td expression was increased. Although the Td expression pattern was similar to RELMa in the heterozygote mice, the fold induction of Td was much lower than that of RELMa. This suggests differences in PCR efficiency, mRNA stability, or that deletion of RELMa has feedback consequences on the RELMa promoter and gene expression. We investigated if these potential differences in RNA levels were also observed at the protein level by flow cytometry. Intracellular RELMa and Td protein was evaluated by flow cytometry of peritoneal cells recovered from PBS or IL-4 complex (IL4c) injected mice ( Figure 1D and Figure S1A). As expected, the main cellular sources of RELMa protein following IL4c injection were the small and large peritoneal macrophages (SPM and LPM) with >95% expression of RELMa in Ra +/+ and Ra Td/+ . In the Ra Td/+ heterozygote mice, Td protein was induced by IL-4 with 50% expression in LPM and 85% expression in SPM. We also observed IL-4c induced expression of RELMa and Td by eosinophils and CD5 + B1 cells ( Figure 1D and Figure S1B). Finally, we examined RELMa and Td expression in other organs such as the visceral fat and the spleen ( Figure S2A), where we observed some IL-4 induced RELMa or Td expression by macrophages, but this was much lower in magnitude compared to the peritoneal cells.
To evaluate heterogeneity in serosal macrophage populations, we generated t-SNE plots on flow cytometry data from IL-4ctreated Arginase-YFP/Ra dual reporter mice ( Figure 1E). The main subsets observed were LPM (red), SPM (blue) and B1 cells (cyan). When comparing heterozygote Ra Td/+ Yarg +/+ and Ra Td/Td Yarg +/+ , RELMa was expressed in SPM and LPM of heterozygote Ra Td/+ but not in homozygote Ra Td/Td mice. Ra Td/Td mice showed instead Td protein expression with similar expression pattern to RELMa. While SPM were a homogenous population with high level RELMa (or Td) expression, LPM exhibited more heterogeneity with mid and high level RELMa-expressing subsets (green vs yellow/red) ( Figure 1F). In contrast, Arginase was more homogeneously expressed in both SPM and LPM (yellow/red). Together, these data validate effective RELMa deletion and replacement with TdTomato and indicate potential heterogeneity of RELMa compared to Arginase expression in the LPM. We also demonstrate that Ra Td/+ heterozygote mice have robust Td and RELMa protein expression, supporting the utility of this transgenic mouse model to track RELMa expression.

RELMa Td/Td Mice Suffer From Increased IL-4c Induced Pathology
Serosal macrophages that reside in the peritoneal cavity have important surveillance roles as sentinels for pathogen infections, but also regulate inflammation and can migrate to visceral organs to mediate repair (37). Peritoneal macrophages are main cellular sources with up to 100% RELMa expression following IL-4c injection, however, the function of RELMa in the peritoneal cavity has not been investigated. Wild-type (Ra +/+ ) or RELMadeficient (Ra Td/Td ) mice were injected with PBS or IL-4c. IL-4c injection led to significantly increased RELMa protein in the serum and RELMa mRNA in the peritoneal cells of Ra +/+ mice, while Td mRNA was significantly elevated in peritoneal cells of Ra Td/Td mice ( Figure 2A). As previously reported (14), IL-4c caused increased peritoneal cell numbers in Ra +/+ mice, but peritoneal inflammation was exacerbated in Ra Td/Td mice ( Figure 2B). Flow cytometric analysis revealed that LPM were the main cell-type affected by RELMa deficiency ( Figure 2C). In the Ra +/+ mice, peritoneal B cell numbers were significantly decreased by IL4c treatment (Figure 2C), and further subsetting into CD5 + B1 cells and CD23 + B2 cells revealed that the decrease was significant in B2 cells ( Figure S2B). In contrast, neither B1 nor B2 cells were reduced by IL-4c in Ra td/Td mice, and B1 cells were significantly higher in IL-4c treated Ra Td/Td mice compared to IL-4c treated Ra +/+ mice ( Figure S2B), suggesting that RELMa is downstream of IL-4c mediated reduction in B cells. Other peritoneal cell subsets were not affected by IL-4c treatment nor RELMa deficiency. IL-4c induces significant LPM proliferation, therefore we evaluated Ki67 expression as a marker for proliferation. There was a significant increase in Ki67 positive LPM and SPM in IL-4c injected Ra Td/Td mice but no changes in B cells ( Figure 2D). RELMa-deficient mice also exhibited IL-4 induced splenomegaly, which was more severe than observed in wild-type mice ( Figures 2E, F). This suggested an exacerbated response in RELMa deficiency, similar to the macrophage activation syndrome caused by sustained IL-4 exposure (38). Proinflammatory cytokine measurement in the serum revealed that Ra Td/Td mice had increased circulating cytokines compared to Ra +/+ mice, with significant increases in IL1a under homeostatic conditions, and increased TNFa, IFNg and IL-6 following IL-4 treatment, but no changes in circulating type 2 cytokine IL-5 ( Figure 2G). We also performed the same cytokine bead array analysis on peritoneal lavage fluid but did not observe detectable cytokine levels. Together, these data reveal that RELMa critically mitigates IL-4-induced inflammatory effects including LPM and SPM proliferation, splenomegaly and systemic proinflammatory cytokine expression. To identify mechanisms underlying RELMa regulation of peritoneal macrophage responses, gene expression analysis was performed in F4/80 + CD11b + peritoneal macrophages sorted from PBS or IL-4c treated Ra +/+ or Ra Td/Td mice, using the Nanostring myeloid immunity panel (734 genes) ( Figure 3A). Principal component analysis (PCA) demonstrated clustering according to genotype and treatment, with IL-4 treatment driving the greatest transcriptional differences regardless of genotype ( Figure 3B). Out of four mice, macrophages from one IL-4c-treated Ra +/+ mice appeared as an outlier and clustered with the PBS-treated group ( Figure 3B, red circle). Retrospective analysis revealed that this mouse had less RELMa in the PEC fluid, and lower peritoneal cell numbers likely because of ineffective IL-4c delivery ( Figure S3A), therefore this sample was removed from gene expression analyses. Investigation of the most differentially expressed genes indicated that Chil3/4 (Ym1/Ym2) and Rnase2a (Ear11) were highly upregulated by IL-4 for both mouse genotypes ( Figures 3C, D). As signature M2 macrophage genes, RELMa and Ym1/2 are reported to have equivalent expression patterns, but Ym1 can also promote RELMa expression (24). Ear11 is an eosinophil-associated ribonuclease that is secreted by M2 macrophages and promotes neutrophil chemotaxis (39). Consistent with IL-4 induced resident macrophage proliferation, genes associated with the cell cycle (Top2a, Cdc20, Kif20a, Ccnb2) were upregulated. Consistent with an anti-inflammatory function for M2 macrophages, both Ra +/+ and Ra Td/Td macrophages from IL-4c treated mice had reduced expression of genes associated with chemotaxis (Cxcl13, Cxcl14, Cxcl2), complement responses (C1qa, C1qb, C1qc) and innate immune activation (Birc3, CD80, CD86) (Figures 3C, D, G). Unexpectedly, the dual-specificity protein phosphatases (Dusp1, Dusp6), and Apoe were also suppressed by IL-4, although these have proposed anti-inflammatory and repair functions (40). These expression patterns likely reflect in vivo macrophage plasticity and the unique response of resident peritoneal macrophages to repeated treatment with IL-4, which may lead to negative feedback pathways for type 2 cytokine signaling. Overall, these genes were similarly induced or suppressed by IL-4 in both Ra +/+ and Ra Td/Td mice, suggesting that these resident M2 macrophage activation programs occur even in the absence of RELMa. We next evaluated RELMa-regulated genes ( Figures 3E-G). The most consistently upregulated genes in PBS or IL-4-induced Ra Td/Td macrophages were MHC class II genes associated with antigen presentation (H2 genes, CD74), suggesting enhanced antigen presentation function by macrophages in RELMadeficient mice even in homeostatic conditions. RELMadeficient macrophages also had increased expression of chemokine/chemokine receptors (Ccl6, Cxcl13, Cxcl16) ( Figure 4A). Cxcl13, involved in B1 cell maintenance (41), was the most upregulated gene in IL-4c induced Ra Td/Td macrophages compared to Ra +/+ macrophages, consistent with the increased B1 cell numbers in RELMa-deficient mice. Conversely Dusp2, which negatively regulates cell proliferation (42,43), was the most downregulated in the Ra Td/Td macrophages ( Figure 4A), consistent with their enhanced proliferation. Advanced pathway analysis (35) was performed to determine functional pathways that were significantly altered by RELMa following IL-4 treatment ( Figure 4B). Consistent with upregulation of genes associated with macrophage hyperactivation, functional pathways that were significantly induced in RELMa-deficient macrophages involved enhanced T cell responses (Th1 activation, T-cell activation, antigen presentation). RELMa-deficient macrophages also induced genes associated with cytokine and growth factor signaling (Pdgfb, Jak3, Syk), which may have contributed to their increased expansion in response to IL-4 ( Figure 4C). This increased macrophage proliferation and frequency in Ra Td/Td mice may therefore result from dual effects of increased growth factor expression and responsiveness, and decreased expression of downregulatory checkpoints, such as Dusp2 and Batf ( Figure 4C). Ra Td/Td macrophages showed a reduction in genes associated with differentiation of myeloid cells (Mafb, Cebpa, Laptm5), which may reflect the enhanced proliferation rather than differentiation or maturation of these macrophages in the absence of RELMa. Genes associated with angiogenesis (Fn1, Pdgfb) (44), phagocytosis (MerTK, Timd4, MHCII, C1q, CD16/32, Rab20 and Anxa1) (45)(46)(47), and the scavenger receptors (Marco, CD163) were increased in Ra Td/Td macrophages ( Figure 4C). This was consistent with the IL-4 treated Ra Td/Td mice exhibiting characteristics of macrophage activation syndrome, associated with splenomegaly and erythrophagocytosis (38). To further analyze and validate some of these genes and their association with RELMa at the single cell and protein level, we use t-SNE mapping of flow cytometry data of peritoneal cells ( Figure 4D). The t-SNE plots indicated the presence of a small subset that shared SPM and LPM characteristics (black arrow). This subset had the highest expression of CD163 and MHC class II, and coexpressed RELMa in the Ra +/+ mice. Further, it was increased by four-fold in the Ra Td/Td mice, suggesting that RELMa expression by this subset may provide an autocrine negative feedback to limit its own expansion. Consistent with the Nanostring data, the MHCII hi expressing subsets ( Figure 4D, red) were expanded in the Ra Td/Td macrophages, especially in the SPM and SPM->LPM subsets, consistent with the significantly increased MHCII MFI in SPM but not LPM in Ra Td/Td mice ( Figure S2C). however, there was heterogeneous distribution in LPM reflecting two functionally distinct LPM subsets in response to IL-4. Anti-CD163 surface staining confirmed the Nanostring data that Ra Td/Td macrophages had significantly elevated CD163 expression in the SPM, LPM and the SPM to LPM cell subset, and was co-expressed with the M2 macrophage marker CD206 ( Figure 4D). CD163 is a scavenger receptor for hemoglobin and is increased in M2 macrophages associated with hemophagocytic syndromes (48). Combined with the increased expression of genes associated with phagocytosis and scavenger functions, our observations that CD163 + M2 macrophages are significantly expanded in Ra Td/Td mice ( Figure 4E) point to a causal link between enhanced macrophage scavenging and the exacerbated IL-4 induced inflammation and splenomegaly in Ra Td/Td mice (38,(49)(50)(51).

RELMa-Expressing M2 Macrophages Support Regulatory T Cell Responses
Our in vivo data suggests that RELMa mice suffer from increased proinflammatory cytokine expression that is associated with enhanced macrophage activation, including increased expression of genes involved in T cell activation. We therefore investigated if macrophage-intrinsic RELMa dampens proinflammatory T cell responses using in vitro co-culture of peritoneal macrophages with splenocytes. Resident peritoneal macrophages from naive Ra +/+ or Ra Td/Td mice were recovered and activated in vitro with IL-4, leading to significant RELMa secretion by Ra +/+ macrophages ( Figure 5A). The macrophages were then recovered and co-cultured with CFSE-labeled splenocytes activated with anti-CD3. After 3 days of co-culture, only modest proliferation of effector CD4 T cells (CD4 + CD25 -) was observed (~15%) (Figures 5B, C), although robust proliferation was observed by day 6 (~70%) ( Figure S3C). Although there were no differences in effector T cell proliferation, CD4 + CD25 + Foxp3 + regulatory T cells (Treg) exhibited robust proliferation (60-80%), which was significantly higher when co-cultured with IL-4 treated Ra +/+ macrophages compared to PBS treated Ra +/+ macrophages ( Figure 5D). PBS-treated Ra Td/Td macrophages supported equivalent Treg proliferation compared to PBS Ra +/+ macrophages, however IL-4 treated Ra Td/Td macrophages were unable to enhance Treg proliferation ( Figure 5D).
We evaluated the downstream effects of macrophage-Treg interaction by quantifying cytokine secretion. Macrophages or splenocytes cultured alone did not produce cytokines ( Figure 5E and data not shown). Both PBS or IL-4 treated Ra +/+ macrophagessplenocyte co-cultures resulted in robust and equivalent secretion of IL-10, GM-CSF, and IFNg, while IL-4-treated macrophages promoted enhanced secretion of IL-1a, MCP-1, IL-6. Co-cultures with PBS-treated Ra Td/Td macrophages induced equivalent cytokine secretion to PBS-treated Ra +/+ macrophages, however, IL-4-treated Ra Td/Td macrophages were unable to promote cytokines associated with Treg differentiation and function (IL-10 and GM-CSF). Instead, IL-4-treated Ra Td/Td macrophages promoted secretion of MCP-1, IL-6. We also observed a reduction in IFNg and IL-1a secretion in co-cultures with IL-4 treated Ra Td/Td macrophages. Given that the co-cultures consisted of macrophages and splenocytes, the cellular source of the cytokines is unclear. Since the splenocytes are treated with anti-CD3, we conclude that most of the cytokines detected are directly from T cells, or indirectly from T cells activating other splenocytes or the peritoneal macrophages to produce cytokines.
We also investigated in vivo-derived M2 macrophages by IL-4c intraperitoneal injection, followed by recovery and purification of F4/80 + macrophages at day 1, and co-culture with anti-CD3 stimulated splenocytes ( Figure 5F). Similar to the in vitro-derived M2 macrophages, co-culture with Ra Td/Td macrophages led to significantly reduced Treg proliferation compared to Ra +/+ macrophages, with no significant effect on effector T cells. The Ra Td/Td co-cultures also had significantly reduced IL-2 levels compared with the Ra +/+ macrophage co-cultures ( Figure 5G), which may explain the reduced Treg proliferation. Together, this data suggests that macrophage-derived RELMa promotes Treg responses and suppresses myeloid expression of chemokines and proinflammatory cytokines, but has a mixed effect on T cell polarization and inflammasome activation.

Dysregulated Splenic T Cell Responses and Reduced Regulatory T Cells in RELMa-Deficient Mice
Based on the co-culture results that demonstrated a direct effect of macrophage-derived RELMa in supporting Treg responses, we sought to determine the in vivo relevance of this novel regulatory function for RELMa. We therefore evaluated peritoneal macrophage and splenic T cell responses in PBS or IL-4c-treated Ra +/+ or Ra Td/Td mice. We observed significantly increased CD25 expression in the Ra Td/Td LPM and SPM ( Figure 6A), which may provide one mechanism for limiting IL-2 availability to Tregs (52). t-SNE mapping showed that most IL-4c induced Ra Td/Td SPM and a small subset of Ra Td/Td LPM expressed CD25, compared to low expression in IL-4c induced Ra +/+ macrophages ( Figure 6B). To validate the in vitro finding of impaired Treg responses in the absence of RELMa, we quantified Treg frequencies in the spleens of PBS or IL-4c-treated Ra +/+ and Ra Td/Td mice. Immunofluorescent analysis of the periarteriolar lymphoid sheath of the spleen confirmed that IL4c-treated Ra Td/Td mice had significantly reduced Foxp3 + cells (Figures 6C, D). Additionally, IL-4c treatment led to detectable RELMa protein expression in Ra +/+ mice and Td protein in Ra Td/Td mice ( Figure 6C), suggesting local effects of RELMa on the spleen. Flow cytometry analysis of the spleen also revealed significant reductions in CD4 + CD25 + Tregs in IL4c-treated Ra Td/Td mice compared to Ra +/+ mice (Figures 6E, F).
We next evaluated if the Treg deficiency in Ra Td/Td mice was associated with dysregulated T cell polarization in the spleen. Anti-CD3 stimulation of splenocytes from IL-4c-treated Ra Td/Td mice led to significantly increased secretion of IL-17A, TNFa and IL-1a compared to splenocytes from the counterpart Ra +/+ mice ( Figure 6G). Combined, these in vitro and in vivo data reveal a previously unappreciated role for peritoneal macrophage-derived RELMa in mitigating IL-4 induced infl ammation and immunopathology through promoting Treg responses and limiting proinflammatory macrophage and T cell responses.

DISCUSSION
As critical sentinels of the peritoneal cavity and visceral organs, the biology of peritoneal macrophages is increasingly being investigated (2,(53)(54)(55)(56). These studies highlight the complexity and importance of these cells in response to infection and inflammation (55,57), but also reveal their role in immune homeostasis (1,2,58,59). Peritoneal macrophages follow similar activation pathways to other macrophage lineages, where M1 macrophages activated by IFNg and TNFa have enhanced microbicidal or tumoricidal capacity and secrete high levels of pro-inflammatory cytokines and mediators (60), while IL-4 activated M2 macrophages reduce inflammation and contribute to tissue repair through secretion of IL-10 and TGF-b (61,62). Although M2 macrophages have antiinflammatory roles, dysregulation of these signaling pathways also induce inflammation and immunopathology (38), which we investigated using the in vivo model of IL-4c induced peritonitis. RELMa is a signature marker of small peritoneal macrophages under homeostatic conditions and is highly expressed by small and large peritoneal macrophages in a type 2 cytokine environment. However, the potential contribution of RELMa to peritoneal macrophage activation, function, or effects on other immune or resident cells are unknown. We addressed this question by generating RELMa transgenic mice and found that RELMa expressed by peritoneal macrophages acts back to limit macrophage proliferation and activation. Macrophage-derived RELMa was also critical to support regulatory T cell proliferation and function. Genetically deficient RELMa mice have been previously investigated (28,29,31), and one study generated a RELMacre recombinase mouse line that enabled fate mapping of RELMa-expressing cells and diphtheria toxin-induced deletion of these cells (23). In the fate mapping RELMa-cre mice, any cell in which the Retnla promoter has been active at any time, will have constitutive reporter protein expression throughout its lifespan, therefore potentially overrepresenting RELMa expression. In contrast, the reporter mouse model described here can reflect temporal changes in Retnla promoter activity. Here, we validate its utility as a faithful reporter by side-by-side analysis of TdTomato reporter and RELMa mRNA and protein expression. Compared to the RELMa-cre mice, or other studies in helminth infection, we found that peritoneal macrophages were the dominant source of RELMa, while eosinophils and B1 cells only expressed modest levels of RELMa in response to IL-4. Our data is consistent with RNA-seq and Immgen datasets that evaluate naïve immune cell subsets (13), where small peritoneal macrophages are the highest RELMa expressors. Compared to other mouse models, these mice offer the potential to specifically delete RELMa within individual cells. Furthermore, this alleviates the need for diphtheria toxin, that causes apoptosis and can have pathologic consequences independent of RELMa function. In this study, we validated our transgenic mice by crossing them to a universal cre-deleter mouse line, however, this transgenic mouse model provides the valuable opportunity in future studies to delete RELMa in specific cell-types. Consistent with other studies demonstrating a protective and anti-inflammatory role for RELMa, we show that RELMa is only expressed in naïve small peritoneal macrophage but is expressed by small and large peritoneal macrophages in IL-4-induced peritonitis. Within the peritoneal cavity, the main cell target of RELMa was the large peritoneal macrophage (LPM), where RELMa limited LPM proliferation and activation. This suggests that the same cell-type that produces RELMa is also its target, suggesting a macrophageintrinsic negative feedback loop. Since IL-4 drives significant expansion of peritoneal resident macrophages, it may be important for immune homeostasis and energy conservation to have internal feedback mechanisms, such as RELMa, to keep this process in-check. For instance, sustained IL-4 exposure leads to immunopathology such as the macrophage activation syndrome, where splenomegaly is observed (38). The treatment regime in our studies involved only two IL-4c injections, however, Ra Td/Td mice had already begun to exhibit immunopathologic features such as splenomegaly.
Although gene expression analysis was performed on bulksorted peritoneal macrophages and did not distinguish monocyte-derived SPM from resident LPM, RELMa deficiency resulted in a heterogeneous macrophage phenotype with SPM and LPM markers. These included increased expression in RELMadeficient macrophages of LPM-specific genes such as Timd4 and Cxcl13, but also SPM-associated genes MHCII, CD62L (Sell), CD38 and CD74 (71,72). IL-4c-induced peritonitis has been previously shown to be caused by resident LPM proliferation rather than the recruitment of blood monocytes (17,73). In our studies both Ra Td/Td SPM and LPM showed evidence of increased proliferative capacity with elevated Ki67 expression compared to Ra +/+ macrophages, yet we did not observe significantly increased SPM numbers. It is possible that in the RELMa -/environment, SPM were transitioning to LPM, as has been reported in inflammatory environments (71)(72)(73). Indeed, the absence of RELMa led to increased circulating inflammatory cytokines including TNFa, IFNg, IL-6 and IL-1a, and exacerbated splenomegaly. Gene expression analysis revealed increased genes associated with growth factor signaling and angiogenesis (e.g. Pdgfb, Ncf2 and Fn1) in the RELMa-deficient macrophages, which could have contributed to the splenomegaly.
The main regulatory effects of RELMa in limiting inflammation and immunopathology were observed following IL-4 treatment, however, MHCII genes (H2-Aa, H2-Ab1 and H2-DMa) were consistently elevated in the RELMa-deficient macrophages in both PBS and IL-4c treatment, suggesting a potential effect of RELMa on antigen presentation in a homeostatic or type 2 cytokine environment. To investigate the role of macrophageintrinsic RELMa in T cell responses, we performed splenocyte co-cultures with peritoneal macrophages from wild-type or RELMa-deficient mice. RELMa deficiency had no significant effect on effector T cell responses, however, was unable to support optimal regulatory T cell proliferation. The direct mechanism underlying this defect is unclear, however, cytokine quantification revealed decreased Treg-associated cytokines GM-CSF (74) and IL-10, and conversely increased IL-6 in RELMa-deficient macrophage co-cultures. These co-culture findings were supported by the in vivo studies, where there was reduced Treg frequency in the spleen, but enhanced Th17 cell responses. RELMa-deficient macrophages had increased expression of the IL-2R (CD25), suggesting that they may limit IL-2 availability to the Tregs (75,76), which was supported by our finding that IL-2 levels were significantly reduced in RELMadeficient macrophage co-cultures. However, further experiments are needed to functionally link CD25 expression with IL-2 consumption by RELMa-deficient macrophages. Also, future investigation of the Tregs is warranted, such as their ability to suppress naïve T cells, and how their function is altered by RELMa. Immunofluorescent staining validated Treg reduction in the spleen, which may have contributed to the splenomegaly by removal of this regulatory brake. Macrophages in the spleen express RELMa, therefore splenic macrophage function may be similar to what was observed for the peritoneal cavity macrophages. An alternative possibility is that peritoneal macrophages migrate to the spleen, as prior studies observed peritoneal macrophage migration to other organs such as the liver in response to injury (2). Previous studies used bone marrow-derived macrophages or dendritic cells and in vivo helminth infection to address RELMa function in T cell polarization (28)(29)(30). Findings from these studies revealed many effects of RELMa on T cells, such as limiting Th2 cell polarization and promoting T cell-derived IL-10. Our studies support an immune regulatory role for RELMa on Tregs. However, we did not observe any differences in Th2 cell polarization, potentially because we interrogated the effect of RELMa on IL-4-induced responses, compared to the more complex outcomes and regulatory networks in helminth infection. Here, we further demonstrate the in vivo significance of resident peritoneal macrophages, which express significantly higher levels of RELMa than in vitro bone marrowderived cells. We also identify a targeted effect of RELMa on promoting Treg proliferation with functional consequences to limit spleen immunopathology. Overall, these studies identify dual effects of macrophage-intrinsic RELMa in limiting macrophage activation while supporting Treg responses with the overall effect of limiting type 2 cytokine-mediated immunopathology. Investigation of this macrophage-Treg axis, and how it is influenced by RELMa, will be an important future direction to assess the biological significance of this interaction beyond IL-4c injection. Specifically, this immune regulatory role for RELMa in the peritoneal cavity may critically influence the outcome of type 2 cytokine-biased diseases such as helminth infection, injury and repair to visceral organs (77)(78)(79)(80), but conversely may have impact in other settings where peritoneal macrophages and Tregs are detrimental such as in cancer metastases (81,82).

DATA AVAILABILITY STATEMENT
The data has been uploaded to NCBI -accession number is GSE174606.

ETHICS STATEMENT
The animal study was reviewed and approved by University of California Riverside Institutional Animal Care and Use Committee.

AUTHOR CONTRIBUTIONS
MN and JL conceptualized the study. JL and NL developed the methodology. JL, SK, and NL performed the investigation. MN and JL performed the formal analysis. MN and JL wrote the article. SK and DC edited the manuscript. MN and DC supervised the study. All authors contributed to the article and approved the submitted version.