MicroRNA-155 Protects Group 2 Innate Lymphoid Cells From Apoptosis to Promote Type-2 Immunity

Group-2 innate lymphoid cells (ILC2) play critical roles in the initiation and maintenance of type-2 immune responses, predominantly through their production of the type-2 cytokines IL-5, IL-9, and IL-13. ILC2 are essential for the efficient elimination of helminth parasites, but also contribute to the detrimental type-2 immune responses that underlie diseases such as asthma and allergy. While several transcription factors have been identified that regulate the development and function of ILC2, less is known about the post-transcriptional mechanisms that regulate these processes. We identified micro-RNAs (miRNAs) that are co-ordinately regulated in ILC2 from mice exposed to two different stimuli, namely IL-33 “alarmin” administration or Nippostrongylus brasiliensis parasitic worm infection. miR-155 is upregulated in ILC2 in response to both stimuli and miR-155−/− mice had impaired IL-33-driven ILC2 responses. Using mixed bone marrow chimeras, we demonstrate that this deficit is intrinsic to ILC2 and that miR-155 protects ILC2 from apoptosis, while having little impact on ILC2 proliferation or cytokine production. These data reveal a subset of miRNAs that are regulated upon ILC2 activation and establish a specific role for miR-155 in regulating ILC2 survival following activation.

ILC2 originate from common lymphoid progenitors (CLP) and several key transcription factors have been identified that regulate their development and function (11). Development of all ILCs requires the combined actions of several key factors, including Id2, Gata3, PLZF, and TCF-1 (12). Lineage-specific factors, such as RORα, Bcl11b, and high levels of Gata3, are subsequently required to specify the ILC2 lineage specifically.
Although there has been significant focus on the transcription factors that regulate ILC2 function, relatively little is known about the post-transcriptional mechanisms that govern ILC2 development and function. MicroRNAs (miRNAs) are important regulators of cell fate decisions and act to co-ordinately repress the expression of multiple genes in order to drive cellular reprogramming. Indeed, miRNAs have been shown to play critical roles in Th2 cell development and function and perturbations in the expression of particular miRNAs are associated with allergy and asthma (13,14). For example, miR-24 and miR-27 cooperate to suppress Th2 cell differentiation, serving to repress expression of the canonical Th2 transcription factor GATA3, as well as other critical Th2 cell regulators such as Ikzf1, and IL-4 in mice (15). Conversely, miRNA (miR)-19a promotes Th2 cell development through its repression of negative regulators of TCR signaling (Pten) and type-2 cytokine production (Socs1 and Tnfaip3) and is reported to be upregulated in T cells from asthmatic airways (16). Recently, the miR-17∼92 cluster that includes miR-19a was also found to promote IL-5 and IL-13 production from ILC2 (17).
A Th2 cell-intrinsic role has also been identified for miR-155, which is contained within exon 2 of the non-coding RNA, B cell integration cluster (bic). Inhibitors of miR-155 reduced IL-13 and IL-5 expression by Th2 cells, suggesting that miR-155 contributes to Th2 effector function, and miR-155-deficient mice exhibited significantly reduced airway inflammation and eosinophilia in response to house dust mite challenge or a model of ovalbumininduced allergic airway inflammation (18,19). A number of putative miR-155 targets have been identified in Th2 cells including genes encoding the sphingosine-1-phospate receptor 1 (S1PR1) (19), which may control lymphocyte egress from lymphoid tissue, the transcription factor PU.1 (18), a negative regulator of type-2 cytokine production, and metabolic regulators such as the mTOR binding partner, RICTOR (19). Recently, miR-155 has also been reported to regulate ILC2 and it was observed that miR-155-deficient mice have reduced numbers of ILC2 in response to ovalbumin-induced lung inflammation (20). Furthermore, ILC2 from miR-155 −/− mice exhibited impaired proliferation and IL-13 production in response to IL-33, though this study did not fully address whether this reflected an ILC2intrinsic role for miR-155, or an effect of miR-155-dependent stromal or T cell-dependent factors (20). Indeed, allergeninduced IL-33 induction in the lung was significantly impaired in miR-155 −/− animals, suggesting pleiotropic roles for this miRNA (20).
To identify miRNAs that regulate ILC2 development and function, we performed comprehensive microarray analysis, with PCR verification, to identify miRNA expression in resting ILC2 from naïve mice, and activated ILC2 isolated from mice administered exogenous IL-33, or infected with the parasitic helminth Nippostrongylus brasiliensis. A subset of miRNAs was differentially expressed in both IL-33 and N. brasiliensis stimulated ILC2, while other miRNAs were regulated according to the particular stimulus used. Among those co-ordinately regulated by both IL-33 and N. brasiliensis infection were immunoregulatory miRNAs, including miR-155. Focussing on miR-155, due to its reported roles in type-2 immunity, we used mixed bone marrow (BM) chimeras and in vitro proliferation and survival assays to demonstrate that ILC2intrinsic expression of miR-155 is required to protect ILC2 from apoptosis.

MATERIALS AND METHODS
Mice bic/miR155-deficient mice (Mir155 tm1.1Brd ) (21) and Rora fl/fl x Il7ra Cre mice have been described previously (1) and CD45.1 + transgenic (B6SJL) mice were produced in-house. All mouse experiments were undertaken with the approval of the UK Home Office. Mice were kept at a specific pathogen-free facility.

Murine Models
Three murine models were used to study ILC2 responses in the presence or absence of miR-155. For IL-33 exposures mice were injected intraperitoneally with 1 µg of recombinant mouse IL-33 in PBS for 3 consecutive days. For IL-2/IL-25 expansion of ILC2 in vivo, mice were injected with 0.5 µg IL-2, 2.5 µg anti-IL2 and 0.5 µg of IL-25 in PBS for 3 consecutive days (1). Mice were euthanized on day 4. For N. brasiliensis infection, mice were administered 500 viable third-stage N. brasiliensis larvae subcutaneously and culled on day 5. Infection was confirmed by enumerating adult worm burden in the small intestine.

Antibodies
The following antibodies were used: FITC-conjugated anti-CD3 RNA was isolated using trizol and the Qiagen RNeasy kits. Samples were analyzed using the Agilent RNA 6000 Pico Chip and 2100 Bioanalyser. cDNA for qPCR was synthesized using the Quantitec Reverse Transcription Kit and the Taqman microRNA assay (mmu-miR-155 and U6) according to the manufacturers' specifications.

MicroRNA Screen
ILC2 (CD3 − CD4 − Lin − ICOS + cells) were isolated from naïve, IL-33 or N. brasiliensis treated mice using fluorescent cell sorting (Sony iCyt Synergy cell sorter) and stored after the addition of TRIzol LS. RNA was extracted using chloroform phase separation, followed by DNase treatment and further purification using Qiagen RNeasy micro kits (performing washes with 100% ethanol to preserve short RNA sequences). RNA integrity was checked using a RNA6000 Pico Kit on an Agilent 2100 Bioanalyzer. Labeling, microarray hybridization, washing, scanning and feature extraction were performed as described previously (22). As up to 150 mice had to be used to extract sufficient microRNA for microarray analysis of naïve mice, only a single measurement was taken due to cost and ethical considerations. Microarrays were scanned on an Agilent Microarray Scanner (G2565CA) using miRNA_107_Sep09 scanning protocol (Agilent). Spot intensities were then extracted using Agilent's Feature Extraction software v10.7.3. Microarray results were compared to qPCR, obtained using Exiqon's miRCURY LNA Universal RT microRNA PCR Mouse&Rat plates I and II V3.M.

Retroviral Construct Preparation
A pSiren-RetroQ retroviral construct (Clontech) was modified by replacing the puromycin resistance gene with an mCherryexpressing gene. Two double stranded oligonucleotides with homology to the entire region of mmu-miR-155-5p and containing BamHI/EcoRI overhangs and an internal hybridisation region, were annealed and ligated downstream of the U6 promoter of the pSiren-retroQ mCherry retroviral vector.

Construction and Collection of the Recombinant Retrovirus
HEK-293T cells maintained in DMEM/10%FCS/penicillin and streptomycin were co-transfected (5 × 10 6 cells) using polyethylenimine with 10 µg of anti-miR-155 pSiren-retroQ and 15-20 µg of the ecotrophic packaging vector pCL-Eco (Imgenex) and the cells were incubated at 37 • C. After 72 h, the supernatant containing the retroviral particles was collected and filtered through a low protein-binding filter (0.45 µm). The supernatant was centrifuged at 25 k rpm at 4 • C for 2 h using an ultracentrifuge and the supernatant removed. The virus particles were resuspended in a primary cell culture compatible media (RPMI, 10% FCS, P/S) for a 20x concentration and stored at −80 • C.
Apoptosis JC-1 staining was performed according to the manufacturers (ABCAM) instructions. AnnexinV staining was performed using eBioscience PerCP-efluor710 labeled AnnexinV and a fixable viability dye. As controls, mitochondria were depolarized using FCCP or cell death induced using heat shock.

Statistical Analysis
Spot intensities from the Agilent Microarray were obtained with background correction and gProcessedSignal data were analyzed. The minimum measure of variability was calculated and added to each value to remove negative values and data was Log2 transformed. Principal component analysis was performed as well as a two-way ANOVA to assess miRNA fold changes and generate hierarchical clustering heat maps. All miRNA analysis was performed using Partek R Genomics Suite R software, version 7.17 Copyright © ; 2017. Other data were analyzed using Graphpad Prism using t-tests or ANOVA, as indicated for individual experiments.

Differential Regulation of miRNAs in ILC2 Following Activation
To determine the expression of miRNAs in ILC2 following activation, we isolated ILC2 (CD3 − CD4 − Lineage − ICOS + ) from the abdominal lymph nodes of naïve wildtype (WT) mice or WT mice treated with IL-33 or infected with N. brasiliensis (day 5 post-infection). miRNA expression profiles were established using microarray and confirmatory PCR (IL-33 data only, data not shown). Principle component analysis (PCA) is shown in Figure 1A. Many miRNA already identified to be involved with lymphocyte activity were detected in this data set and their relative expression between stimulated and naive samples are shown ( Figure 1B). Data revealed 95 miRNAs in ILC2 that were differentially regulated in response to both IL-33 and N. brasiliensis, with a further 67 unique to IL-33 and 21 unique to N. brasiliensis ( Figure 1C, Supplementary Tables 1-7). Of these, miR-155 was selected for further study as it ranked highly in fold-change (18.9-fold increase) in IL-33-treated mice and has been implicated previously in lymphocyte development and Th2 cell immunity. miRNA differentially expressed between naïve ILC2 and ILC2 following in vivo IL33-stimulation or N. brasiliensis-infection, and T cells upon N. brasiliensisinfection indicated considerable co-incidence of miRNA modulation ( Figure 1C and Supplementary Tables 1-7). Notably 34 miRNA were found to be changed in all stimulated populations ( Figure 1C). These 34 miRNAs included mir-155, mir-146a, and mir-19a (Figure 1C), of which mir-155 was among the most highly expressed in activated cells; intensity data for miR155 are shown for each sample ( Figure 1D).

miR-155 Is Required for ILC2 Expansion Response to IL-25 or IL-33
To study the effect of miR-155 on ILC2 we injected both wildtype (WT) and miR155 −/− mice with IL-33 to expand ILC2 in vivo. ILC2 were isolated from the mesenteric lymph nodes (MLN) and enumerated. As expected, IL-33 increased ILC2 numbers in WT mice. However, there were significantly fewer ILC2 in the MLN of miR-155-deficient mice treated with IL-33 (Figures 2A,B). There were no differences in the expression of ST2 (the IL-33 receptor) between WT and miR155 −/− ILC2 ( Figure 2C). Furthermore, although administration of IL-25 in combination with a complex of IL-2 and anti-IL2 (to maintain the slow release of IL-2 and maximize ILC2 proliferation) also induced an expansion in ILC2 numbers in WT mice, this increase was reduced in the absence of miR-155 ( Figure 2D). Thus, miR-155 is required for the increase in ILC2 numbers in response to IL-25 or IL-33 stimulation.
Haematopoetic Cell miR-155 Is Critical for ILC2 Expansion in Response to IL-33 Early studies suggested that miR-155 is only expressed in haematopoetic cells (23). However, subsequent studies have reported expression of miR-155 in other cell types, such as endothelial cells (24,25), epithelial cells (26,27) and fibroblasts (28). To exclude potential miR-155-related perturbation of epithelium-derived cytokines or other stromal cell factors, we generated bone marrow (BM) chimeric mice to study the effects of miR-155 within the haematopoietic compartment. Lethally irradiated WT CD45.1 + (B6SJL) mice received CD45.2 + BM from WT mice, miR-155-deficient mice, or an equal mix of WT and miR155 −/− BM cells. Following a period of 3 months (to allow reconstitution of donor-derived ILC2 populations), mice were treated with intra-peritoneal IL-33 and the number of donor-derived ILC2 enumerated. In mice that had received miR155 −/− BM only, there were fewer ILC2 in the MLN compared to mice that had received WT BM ( Figure 3A). There were no differences between WT and mixed WT/miR155 −/− BM transplanted mice, indicating that the absence of miR-155 does not have a trans-inhibitory effect. Thus, the deficit in ILC2 numbers in miR-155-deficient mice is dependent on haematopoietic lineages, but not stromal cell signals.

miR-155 Does Not Regulate Numbers of ILC2 Precursors
To determine if miR-155 affected the expansion of ILC2 precursors, we studied the levels of common lymphoid progenitors (CLP), common helper innate lymphoid progenitors (CHILP) and ILC2 in the BM of IL-33-treated chimeric mice. There were no significant differences in BM CLP (Figures 3B,C), CHILP (Figure 3D), or ILC2 ( Figure 3E) Thus, deficits in progenitor and precursor populations do not underlie the reduction in ILC2 numbers observed in the lymph nodes following IL-33 treatment.

miR-155 Increases ILC2 Numbers Through an ILC2 Intrinsic Mechanism
To establish whether the effect of miR-155 was intrinsic to ILC2 themselves, and not the actions of miR-155 in other haematopoetic cells (such as T cells) or indeed miR-155 containing exosomes derived from other haematopoetic cells (29, 30), we generated mixed WT/ILC2-deficient or miR155 −/− /ILC2-deficient BM chimeric mice. ILC2-deficient BM was obtained from Rora flox/flox x Il7ra Cre mice (1). In these mice, the essential ILC2 transcription factor RORα is deleted in cells expressing the IL-7 receptor. This results in a depletion of ILC2, while there are no reported defects in T helper cell function (1,31). Lethally irradiated WT CD45.1 + (B6SJL) recipient mice received either mixed equal proportions of WT/ILC2-deficient, or miR155 −/− /ILC2 deficient BM. In these mice, ILC2 are derived from either WT or miR155 −/− cells, but other cells (stromal as well as at least a proportion of the haematopoetic cells) have unperturbed miR-155 signaling. While WT ILC2 expanded robustly in response to IL-33, this expansion was diminished in the absence of ILC2 intrinsic miR-155 (Figures 4A,B). There were no differences in MLN T-cell (CD4 + - Figure 4C) and B-cells (B220 + - Figure 4D) in the presence or absence of miR-155. These results support an ILC2-intrinsic role for miR-155 in regulating ILC2 expansion in response to IL-33.

miR-155 Is Required for ILC2 Expansion in vitro
To confirm the effects of miR-155 on the expansion of ILC2 that we observed in vivo, we harvested ILC2 from the mesenteric lymph nodes (MLN) of mice treated with IL-2/anti-IL-2/IL-25. Flow cytometrically purified ILC2 were expanded in culture with IL-7 and IL-33, and cell counts performed at regular intervals. As suggested by the in vivo data, we observed lower numbers of ILC2 in the absence of miR-155 ( Figure 5A). To substantiate that this effect was miR-155-specific we transduced miR-155-deficient ILC2 with a retroviral vector over-expressing miR-155 (rescue) or a negative control sequence. Retroviral transduction of miR-155-deficient ILC2 with the negative control miRNA failed to restore ILC2 number to that of control cultures ( Figure 5A). By contrast, following retroviral transduction with a vector over-expressing miR-155, miR155 −/− cells numbers were restored to the level of WT ILC2 (Figure 5A). These findings indicate that the effect of miR-155 in enhancing ILC2 numbers in vivo is replicated in vitro and specific to miR-155.

miR-155 Does Not Affect ILC2 Cytokine Expression
A previous publication reported effects of miR-155 on cytokine production by ILC2 (20). Hence, we examined whether miR-155 regulated IL-13 production by ILC2. After 5 days of culture with IL-7 and IL-33, there were no differences in the percentage of cells staining positive for intracellular IL-13 in the presence or absence of miR-155 ( Figure 5B). In addition, we quantified the levels of IL-13 in supernatants collected from cultured WT and miR155 −/− ILC2 (Figure 5C), and ILC2 transduced with miR-155 knock-down vector (or negative control) using ELISA ( Figure 5D). No differences in the levels of IL-13 were detected in the supernatants. We also quantified intracellular IL-13 expression by ex vivo ILC2 obtained from WT/ILC2deficient, or miR155 −/− /ILC2-deficient chimeric mice treated intra-peritoneally with IL-33. No differences were found in the expression of IL-13 in the presence or absence of miR-155 ( Figure 5E). Taken together these results suggest that miR-155 does not directly regulate IL-13 production by ILC2.

miR-155 Protects Activated ILC2 From Apoptosis
To further study the mechanism by which miR-155 increases numbers of ILC2, we assessed potential differences in proliferation between WT and miR155 −/− ILC2. We enriched ILC2 by negative selection from lymph nodes of IL-33-stimulated mice, and after 1 week in culture with IL-7 and IL-33, ILC2 proliferation was assessed using Ki67 staining. We detected no differences in the proportion of Ki67 + ILC2 harvested from WT and miR155 −/− mice and cultured in vitro (Figures 6A,B).
Since we were unable to obtain evidence for a role for miR-155 in supporting ILC2 proliferation, we assessed whether the observed reduction in miR155 −/− ILC2 numbers compared to WT might be explained by increased ILC2 cell death in the absence of miR-155. WT and miR155 −/− ILC2 were expanded in IL-7 and IL-33. Cells were then either kept in growth medium, or apoptosis was induced using serum starvation for 6 h. We stained ILC2 using Annexin V (to detect externalization of membrane phosphatidylserine as a facet of apoptosis) and viability staining. Even under normal ILC2 culture conditions the absence of miR-155 resulted in more apoptotic cells in the miR155 −/− ILC2 cultures (Figures 6C,D). Furthermore, serum starvation for 6 h potentiated these differences (Figures 6C,D).
To confirm these findings, we assessed mitochondrial outer membrane potential (as measured by JC-1 staining), as an alternative indicator of apoptosis. While there was a trend  toward an increase in apoptosis in miR155 −/− ILC2 cultures in complete medium as compared to WT controls (Figures 6E,F), this increase in apoptosis reached statistical significance after 6 h of serum starvation (Figures 6E,F). Although we also investigated the presence of markers of apoptosis directly ex vivo, we found there were very few apoptotic ILC2 in the LN of IL-33-treated mice, likely because apoptotic cells are rapidly cleared in the tissue. Together, our data indicate that the expression of miR-155 protects ILC2 from apoptosis during activation.

DISCUSSION
ILC2 are key players in type-2 immunity, and an understanding of their regulation and biology offers an opportunity for the development of new therapeutic targets in type-2 immune-mediated diseases such as asthma and allergy. Using comprehensive miRNA microarray and RT-qPCR strategies we identified a group of miRNAs whose expression was regulated in response to ILC2 activation in vivo, following either IL-33 treatment or N. brasiliensis infection, compared to naïve controls. Additional miRNAs were also regulated uniquely in response to IL-33 administration or worm infection, respectively. Of the shared miRNAs several have been associated with the regulation of cell growth and proliferation e.g. miR-93 (32, 33), miR-25 (34), miR-103 (35). Notably, others are known to regulate immune function, including miR-155, mir-146a and let-7a and let-7d (36). Furthermore, our data show good concordance with the ILC2 miRNA expression reported recently by Singh et al. using small RNA-sequencing (17). They went on to show that ILC2 deficient in the miR-17∼92 cluster were less proliferative than wildtype controls, though the rate of apoptosis was not affected. miRNA-19a, located in the miR-17∼92 cluster was also required for optimal IL-5 and IL-13 production by ILC2, and this contributed to allergic inflammation (17).
It has been reported recently that both miR-155 and miR-146a are required for the type-2 response to helminth infection. MiR-146a-deficient mice are unable to clear Trichuris muris infection, associated with a decrease in IL-13-producing T cells and an increase in IL-17-secreting T cells (19). Heligmosomoides polygyrus infection of miR-146a-deficient mice led to a decrease in the proportion of IL-5-producing T cells with an increase in IL-17 and IFN-γ producing T cells, suggesting that miR-146a was responsible for polarizing T cells to a Th2 profile. Furthermore, miR-155-deficient mice were also unable to mount a Th2 response when infected with H. polygyrus and miR-155 was upregulated in Th2 cells as well as the cells around the intestinal lesions during an H. polygyrus infection (19). MiR-155-deficient mice also had significantly lower IL-13 expression and airway mucus in response to house dust mite challenge (19), and impaired responses to an OVA-model of allergic airway inflammation (18). Additionally, miR-155 has been shown to play key roles in other cell types and immune responses. In the absence of miR-155, mice fail to produce high affinity IgG1 and have smaller germinal centers (37)(38)(39). MiR-155 is also crucial in Th17 cells (40), and mice deficient in miR-155 are resistant to experimental autoimmune encephalitis due to reduced production of Th17 and Th1 responses (41). Furthermore, in the absence of miR-155, CD8 T-cell responses against viruses and tumors are impaired (42). Finally, a recent report showed that ILC2 proliferate less readily in the absence of miR-155, however, an ILC2-intrinsic requirement for miR-155 in vivo was not explored and the mechanism by which miR-155 increased ILC2 numbers was not determined (20).
We confirmed that there are fewer ILC2 in the lymph nodes of IL-33-treated miR-155-deficient mice compared to WT controls, and that this is due to an ILC2-intrinsic role for miR-155 in maintaining ILC2 numbers during immune activation by protecting them from apoptosis. While miR-155 is widely expressed in haematopoetic cells, several reports indicate expression in stromal tissues, including epithelial cells. Generating BM chimeric mice allowed us to demonstrate that the effects of miR-155 on ILC2 were dependent on the haematopoetic compartment. Furthermore, by generating mixed BM chimeras using miR155-deficient BM and ILC2-deficient BM from Rora flox/flox x Il7ra Cre mice, we established that the effect of miR-155 on ILC2 numbers is intrinsic to ILC2. This effect was recapitulated in vitro, with fewer miR155 −/− ILC2 present after culture with IL-7 and IL-33 as compared to WT ILC2, and this was rescued by retroviral over-expression of miR155.
Interestingly, there were no differences in BM ILC2 precursors, such as haematopoetic stem cells, CLP or CHILP in the presence or absence of miR-155. Thus, miR-155 did not appear to regulate ILC2 numbers in vivo by increasing proliferation of ILC2 precursors. Also, there were no differences in BM ILC2. Hence, it is unlikely that the increased numbers of ILC2 in peritoneal lymph nodes were caused by increased ILC2 progenitor production.
Notably, although others have reported a defect in miR-155deficient ILC2 proliferation (20), we did not observe differences in the presence of Ki67 + ILC2 cultured in vitro or when analyzed ex vivo, as compared to WT controls. In contrast to Johansson et al. who assessed the percentage of total Ki67 + Lin − cells compared to total lung cells (which does not take into account the differences in ILC2 numbers that are present in the absence of miR-155), we specifically determined the proportion of proliferating ILC2 compared to total ILC2 to account for the differences in total ILC2 numbers in the miR155 −/− mice. Furthermore, as others have also reported effects of miR-155 on cytokine production (21), and indeed on the production of IL-13 by ILC2 (20), we measured IL-13 production by ILC2 in the presence or absence of miR-155. However, we did not find any differences in the ability of ILC2 to produce IL-13 in the presence or absence of miR-155, as measured by ELISA and intracellular cytokine staining. It is not clear why these data differ, though the mouse lines used in the different studies, or the facilities, could contribute. Our study employed the mir-155-deficient mouse strain (Mir155 tm1.1Brd ) generated by Bradley and colleagues (21), whilst Johanssen et al. used the miR155 −/− mice (Mir155 tm1.1Rsky ) from Rajewsky and colleagues (38). Both groups targeted exon 2 in 129Sv-derived embryonic stem cells, removed the neomycin selection cassette, and backcrossed to a C57BL/6 background. However, the Mir155 tm1.1Rsky line also retains a LacZ reporter. An alternative explanation for the differences observed is that our study and that of Johannson et al. investigated ILC2 from different tissues; lymph nodes or lungs, respectively. Consequently, variances may arise from the kinetics of the response and the time-points analyzed.
Since we had confirmed that the effects of miR-155 on ILC2 numbers were preserved in response to IL-25, and that miR-155 did not alter ST2 levels, it is unlikely that miR-155 interferes with IL-33 receptor signaling specifically. However, using two different methods to quantify apoptosis (JC-1 staining as a measure of loss of mitochondrial membrane potential, and annexin V), we identified that miR-155 expression protects stressed ILC2 from undergoing apoptosis, thereby maintaining and increasing ILC2 numbers. Previous studies have linked miR-155 to apoptosis, both in pro-and anti-apoptotic roles. MiR-155 has also been shown to be pro-apoptotic in acute myeloid leukemia (43) and human dendritic cells (44). However, extensive evidence links miR-155 with anti-apoptotic functions, including in immune cells. In pancreatic cancer ductal cells, miR-155 targets tumor protein 53-induced nuclear protein 1, resulting in decreased apoptosis and increased tumorogenic potential (45). Others have suggested that miR-155 targets caspase-3 and Fas-associated protein with death domain (FADD) directly to protect cells from apoptosis (46). In psoriasis, miR-155 is thought to result in increased proliferation and reduced apoptosis by targeting the PTEN signaling pathway (47). FOXO3 is a target of miR-155, and miR-155 inhibits apoptosis by targeting FOXO3 in monocytes (48) and T cells (49). In B cells, JARID2 is targeted by miR-155, inhibiting apoptosis and resulting in germinal center B cell expansion through collaboration with c-Myc (50). To our knowledge our data are the first linking miR-155 to the control of apoptosis in innate lymphoid cells. Interestingly, recent studies have also indicated that miR-155 promotes autophagy to protect cells during stresses such as hypoxia and nutrient deprivation (51,52). Thus, it is possible that in the absence of miR-155 ILC2 have impaired autophagy and that this contributes to their decreased survival. However, further studies will be required to identify miR155 target genes in the regulation of apoptosis in ILC2, and to investigate potential links between miR155 and autophagy-associated cell death (53).
In summary, we present a comprehensive analysis of miRNA changes in ILC2 in response to activation and demonstrate that miR-155 increases ILC2 frequency in response to stimulation by reducing apoptosis. Although we were unable to identify the target(s) of miR-155 that orchestrate control of apoptosis in ILC2, our data provide a novel pathway for the regulation of ILC2 through apoptosis.

AUTHOR CONTRIBUTIONS
MK, SC, AE, and AG designed and performed experiments, analyzed data, and made figures. BR analyzed data. RN and PF provided reagents. AM conceived the study and designed experiments. All authors contributed to writing the paper.

FUNDING
AM is supported by funding from the MRC (U105178805) and Wellcome Trust (100963/Z/13/Z).