c-FLIP is crucial for IL-7/IL-15-dependent NKp46+ ILC development and protection from intestinal inflammation in mice

NKp46+ innate lymphoid cells (ILC) modulate tissue homeostasis and anti-microbial immune responses. ILC development and function are regulated by cytokines such as Interleukin (IL)−7 and IL-15. However, the ILC-intrinsic pathways translating cytokine signals into developmental programs are largely unknown. Here we show that the anti-apoptotic molecule cellular FLICE-like inhibitory protein (c-FLIP) is crucial for the generation of IL-7/IL-15-dependent NKp46+ ILC1, including conventional natural killer (cNK) cells, and ILC3. Cytokine-induced phosphorylation of signal transducer and activator of transcription 5 (STAT5) precedes up-regulation of c-FLIP, which protects developing NKp46+ ILC from TNF-induced apoptosis. NKp46+ ILC-specific inactivation of c-FLIP leads to the loss of all IL-7/IL-15-dependent NKp46+ ILC, thereby inducing early-onset chronic colitis and subsequently microbial dysbiosis; meanwhile, the depletion of cNK, but not NKp46+ ILC1/3, aggravates experimental colitis. In summary, our data demonstrate a non-redundant function of c-FLIP for the generation of NKp46+ ILC, which protect T/B lymphocyte-sufficient mice from intestinal inflammation.

1) The authors conclude that NKp46+ ILC1 as well as cNK cells are protective for DSS colitis but the mechanisms are largely untouched. Does IFNγ play any role in this protective effect? The authors demonstrated the reduction of IL-12p70, MCP-1 and GM-CSF in c-FLIP<flox/flox>NKp46<iCre> mice but cells producing thoses cytokines are not shown. Are ILC1/NK cells major producer of those cytokines or other cells types are activated by ILC1/NK cells to produce those cytokines?
2) In many inflammatory conditions, CD11b+Gr1+ so-called myeloid suppressor cells are induced and play anti-inflammatory role. It is of informative to examine the induction of such population by ILC1/NK cells.
3) The authors useed RAG1-/-IL-7R-/-mice to show in Fig. 2 that cNK cells are not affected by the lack of IL-7 signals but all other ILCs are reduced. The authors used in Fig. 1 IL-15-/-and IL-15R-/mice to show the reduction of cNK cells but they did not perform detailed analysis as Fig. 2. The authors should show the frequency and numbers of NK1.1+ ILC1, T-bet+ ILC1, EOMES+ c NK and RORγt+ ILC3 in RAG1-/-IL-15-/-or RAG1-/-IL-15R-/-mice as well.
4) The authors examined gut microbiota compositions and discussed a possibility that increase of certain bacterial species such as Turicibacter and Bacteroides could lead to inflammation. However, there is no direct evidence to show that these bacteria play major role on the inflammation. It is similarly possible that the changes in bacterial species is just a results of inflammation. What happened if WT and c-FLIP<flox/flox>NKp46<iCre> mice were co-housed before and during DSS chronic colitis induction?
5) The authors showed IFNγ expression by YFP knockin mice in Fig. 7. Are the cells activated or left unstimulated?
Reviewer #2 (ILC differentiation, cytokine)(Remarks to the Author): The authors investigate the maintenance of NKp46+ ILC subsets in mice lacking the Cflar gene in all NKp46+ cells (Ncr1-Cre; Cflar/Flipfl/fl mice). They observe that splenic NK cells were >90% reduced in these mice and the fraction of NK cells undergoing apoptosis was increased whereas the population of bone marrow NK "precursors" was enriched. In the intestine, all NKp46+ ILC (i.e., cNK, ILC1 and NKp46+ ILC3) were reduced in numbers whereas ILC2 were unaffected. Mice lacking c-FLIP in all NKp46+ cells developed more severe acute and chronic DSS colitis with inflammation-induced changes in microbial communities. While deletion of the IL-7R on all NKp46+ cells had no effect on colitis severity, preferential depletion of cNK cells with anti-asialo-GM1 led to enhanced colitis scores. This is an interesting manuscript suggesting that NKp46+ ILC may protect against DSS colitis. The data also extends on previous work (e.g., Klose, Cell 2014) showing that all NKp46+ ILC require IL-15 signaling for maintenance and/or survival. The data is of good quality, but at this stage, it is rather descriptive and needs additional decisive data to better link NKp46+ ILC subsets to DSS-induced colitis.
The following specific issues are raised.
1. I do not understand the title of the paper. The authors show that anti-asialo-GM1 depletion mainly affects cNK cells and leads to more severe colitis. Yet, the title claims that ILC1 are protective in DSS colitis. This is misleading and should be corrected. My understanding of the authors' main message is that cNK cells ameliorate DSS colitis.
2. The data support the view that cNK cells protect against DSS colitis. A previous paper (Hall, Mucosal Immunol 2013) has put forward a similar view. In that paper it was demonstrated that this effect is mediated by NKG2A/neutrophil interactions. NKG2A is not expressed by ILC1 so that this would indeed isolate the colitis-protective effect to cNK cells. The authors should discuss these previous findings and provide more decisive data that cNK cells are protective. This reviewer appreciates that there are currently no mouse models available allowing for selective ILC1 deficiency but gain-of-function experiments by adoptive transfer of wildtype cNK cells, ILC1 or NKp46+ ILC3 into DSS-treated Ncr1-Cre; Cflar/Flipfl/fl mice on a Rag-/-background could be performed.
3. The protective role of cNK cells in DSS colitis should be confirmed using mice lacking all cNK cells, i.e., Eomesfl/fl; Ncr1-Cre mice and in mice lacking NKG2A. 4. The authors show that microbial communities change in Ncr1-Cre; Cflar/Flipfl/fl mice treated with DSS. These findings were not further explored. Would transfer of microbiota from Ncr1-Cre; Cflar/Flipfl/fl mice into wildtype mice treated with DSS worsen pathology. Figure 1D  In addition, colon length is often also supported by photographs of colon/cecum resections from the various mouse strains and treatment groups.

The data in
8. It is pivotal to provide absolute cell numbers for cNK cells, ILC1 and NKp46+ and NKp46-ILC3 after anti-asialo-GM1 treatment.

Reply to Reviewer 1
We would like to thank the reviewer for his/her efforts and helpful comments.

Reply (B):
The relative impact of cytokines on disease modulation can change over time 5 . Based on this we hypothesized that persistent production of a particular cytokine correlates positively with its impact on disease modulation. Initially our cytokine assays were performed during the first regeneration phase at d10 (now Fig.  5G; lower panel). To characterize cytokine production at an earlier time point, we repeated our assays at d3 (new set of data; Fig. 5G; upper panel). Only GM-CSF levels were significantly reduced in NK Δc-FLIP mice at both time points. On the contrary, IL-12p70 and MCP-1 were lower only at d10. We therefore analyzed GM-CSF production by colonic LPLs from DSS-treated NK Δc-FLIP mice and controls. Our results are summarized in Fig. 5H demonstrating that absolute numbers of NKp46 + , but not NKp46 -, GM-CSF-producing ILCs were significantly reduced in NK Δc-FLIP mice. Hence, lower numbers of GM-CSF + ILCs may explain the reduced regenerative potential of DSS-treated NK Δc-FLIP mice. However, we also found GM-CSF production by other cells (e.g. T cells) and therefore cannot determine yet to which extent ILCderived GM-CSF protects mice from DSS-colitis. For this purpose, NKp46 iCre x GM-CSF fl/fl mice would be required. However, their time-consuming, labor-intensive and expensive generation and characterization would be beyond the scope of the current manuscript.

2) Reviewer: "In many inflammatory conditions, CD11b+Gr1+ so-called myeloid suppressor cells are induced and play anti-inflammatory role. It is of informative to examine the induction of such population by ILC1/NK cells."
Reply: In the previous version of our manuscript (Fig. 3J/K), we defined CD11b + Gr1 + cells as neutrophils. However, as pointed out by the reviewer, these markers are also used to characterize myeloid-derived suppressor cells (MDSCs). Shaul and Fridlender referred to this dilemma in 2017 as follows 6  Hence, it appears impossible to discriminate between MDSCs and neutrophils in our experimental system. We therefore modified the text accordingly and discuss different possibilities regarding the identity and function of CD11b + Gr1 + cells in DSS colitis. Fig. 2 Fig. 1 IL-15 Reply: To address this point, we analyzed SI LPLs from IL-15 -/-, IL-15Rα -/and WT mice (Suppl. Figure 1). In accordance with Robinette et al. 9 , we observed a strong reduction of cNKs and ILC1s in IL-15 -/-/IL-15Rα -/mice. The abundance of other ILC subsets was not changed significantly. We used mice on a Rag +/+ background to make these SI LPL data comparable to those obtained with IL-15 -/-/IL-15Rα -/splenocytes ( Fig. 1C-F).

4) Reviewer: "The authors examined gut microbiota compositions and discussed a possibility that increase of certain bacterial species such as Turicibacter and
Bacteroides could lead to inflammation. However, there is no direct evidence to show that these bacteria play major role on the inflammation. It is similarly possible that the changes in bacterial species is just a results of inflammation (A). What happened if WT and c-FLIP<flox/flox>NKp46 mice were co-housed before and during DSS chronic colitis induction?(B)"

Reply (A):
We agree with the reviewer. Based on our results we cannot conclude that alterations of the microbiota drive inflammation. Alternatively, alterations of the microbiota may be a result of inflammation. In the previous manuscript we therefore stated in the discussion: "This may, at least partially, rely on the disease-related control of the commensal microbiota. Of note, the lack of NKp46 + ILCs in NK Δc-FLIP mice did not have a significant impact on the composition of the commensal microbiota prior to DSS colitis. This argues against a direct impact of NKp46 + ILCs on the establishment of the microbiota under homeostatic conditions in T/B-competent mice. Nevertheless, after establishment of chronic colitis the composition of the commensal microbiota changed in an NKp46 + ILC-dependent fashion." To further emphasize the point raised by the reviewer, we included the following sentences to the discussion: "Whether the observed alterations of the commensal microflora are a result and/or a driver of inflammation remains an important question for the future."

Reply (B):
For these experiments control littermates were used. Whenever possible, they shared cages from birth on and throughout the experiments. We modified the text accordingly. On page 9 we state: "To minimize potential cage-specific variations in microbiota composition, NK WT and NK Δc-FLIP mice littermates were, whenever possible, co-housed permanently." Furthermore, on page 10 emphasize: "Notably, genotype-dependent differences in microbiome composition were not visible before induction of DSS colitis ( Fig. 6A and B; ADONIS values)." Fig. 7. Are the cells activated or left unstimulated?" Reply (B): Cells were isolated from untreated reporter mice, surface molecules were stained and samples were isolated immediately by flow cytometry. To clarify this point we modified the legend of Fig. 2 accordingly: "(A) To determine Ifng promoter activity in unstimulated ILC precursors, freshly isolated BM cells from eYFP-transgenic IFN-γ reporter mice were analyzed by flow cytometry."