Maternal group 2 innate lymphoid cells contribute to fetal growth and protection from endotoxin-induced abortion in mice

Reviewer #1 (Public Review):

In this paper, authors used IL-33KO and ILC2KO mice to generate evidence for pregnancy specific enrichment of ILC2s and their unique molecular signatures. They documented that uterine ILC2s are distinct from the ILC2s residing in lung and lymph nodes. They have provided solid evidence that although litter size did not change, but fetuses from ILC2KO mice showed growth restriction. In the absence of ILC2s, LPS injections lead to increased abortion rates and fetal loss suggesting the critical role of ILC2s in protection against infection induced pathology. Most of the results presented in the paper support the conclusion, but in some instances the evidence is indirect. Some clarifications on experimental interpretations are required as below.

• The immunohistology experiments revealed that IL-33 predominately co-localizes with ILC2 in the myometrium, but the staining appears to be diffused throughout the myometrium. It is difficult to pinpoint ILC2 specific IL-33 colocalization. Is the decidual expression of IL-33 largely restricted to ILC2s? Also, IL-33 is produced by a variety of other cell types including endothelial cells, which is difficult to ascertain from the images in Fig 1.
• Authors report that ILC2s are responsible for fetal growth restriction (FGR) noted in the ILC2 KO mice. They attribute FGR to utero-placental abnormalities but the experimental evidence, including spiral arterial remodelling and glucose transporter gene expression are indirect. Is there any compensation from other cell types such as macrophages in the absence of ILC2s as they reported increased dendritic cells, neutrophils, and macrophages in ILC2KO mice. Clarify whether IL-33 levels were consistent between WT and ILC2KO mice. How do these other increased numbers of macrophages, DCs and neutrophils fit in the FGR context besides gene expression changes they captured in DCs and macrophages.
• They captured spiral arterial wall to lumen ratio alterations in ILC2KO mice suggesting sub-optimal vascular changes in ILC2KO mice. They did not find any changes in the uNK cells or IFN-production between WT and ILC2KO. What would be the mechanistic link between ILC2 and spiral arterial vascular changes. They indirectly link it to the IL1B gene expression.
• In the LPS induced abortion in ILC2 KO mice experiments, how do they reconcile the predominant role of macrophages and LPS induced TNF-a in the pathology? They did not find any differences in the gene expression in the LPS induced signature cytokine, TNF-a despite increased numbers of macrophages in ILC2 KO mice. Clarification is required on whether these inflammatory alterations that they captured directly linked to utero-placental insufficiency between WT and ILC2KO. The type 2 cytokines were barely detectable in ILC2KO mice, which likely predispose them for utero-placental alterations.

Reviewer #2 (Public Review):

Balmas et al., continue the previous work from multiple groups that suggested the implication of uterine ILC2s and signals that activate them, i.e., IL-33/ST2 axis, in healthy and complicated pregnancies and move forward to understand further their role. The authors leverage available and appropriate tools to address more specifically the role of ILC2s during pregnancy and endotoxin-induced abortion, namely mouse models of selective ILC2 deficiency (Roraflox/floxIl7raCre/wt) and transcriptomic analysis of the immune response.

The authors demonstrate, and therewith confirm findings by Bartemes et al. (2018), that ILC2 reside in the mouse uterus, depend on IL-33 and expand during pregnancy. Moreover, they show the Il33 expression by CD45- cells of the uterine stroma. What remains unclear is the kinetics of Il33 expression and ILC2 expansion upon gestation and whether the local ILC2 population expands or arrives from the periphery.

Lack of maternal ILC2, in a mouse genetic model, resulted, as expected, in the absence of uILC2 but also in lighter fetuses at term, similar to the phenotype observed in the absence of maternal IL-33. It would be interesting to understand whether the effect of the IL-33 signaling is a direct ILC2 mediated effect, as for example by using the ST2flox/flox mice. Do the fetuses catch up in weight with their WT controls during weaning time? Do they have any long-term cognitive/behavioral impairment?
The authors showcase the impairment in the remodeling of uterine wall vessels in dams lacking ILC2. It remains to be verified whether this is dependent on IL-33 and whether it is a direct effect of ILC2 or ILC2-dependent infiltration of eosinophils. Further, the absence of ILC2 is accompanied by an increase in Il1b in the uterine tissue suggesting that uILC2 contribute to the uterine microenvironment.

The authors perform RNA sequencing analysis on the bulk samples of uterine ILC2, where uILC2 cluster separately from corresponding lung and LN cells and are featured with higher expression of typical ILC2 markers. Somewhat odd, the authors report on the Foxp3/FoxP3 expression among uILC2, however the staining is not very bright and a Treg control as well as biological negative control should be provided. Moreover, FoxP3 is also not expressed among intestinal ILC2 with regulatory function (Wang et al. 2017). I suggest this data panel to be re-evaluated. A scRNA-Seq analysis would probably be more comprehensive in this case, but might be beyond the scope of this publication.

Absence of uILC2 results in the increased numbers of DCs, macrophages and neutrophils in the uterus, an impact which is not visible in the spleen, which is why the authors argue that this is a uterus-restricted phenomenon, although perturbances in the large intestine and lungs could be expected. Moreover, it remains to be investigated whether these effects are restricted to mid-term pregnancy or preserved until term.

Upon establishing the role of uILC2 in maintaining healthy pregnancy, the authors demonstrate a role for uILC2 in endotoxin-mimicked bacterial infection and abortion. An impressive set of data demonstrate that dams that lack uILC2 have a significantly higher fetus resorption rate than WT dams upon LPS challenge. It remains to be understood whether this phenotype is also dependent on IL-33. Finally, mechanistically, using a somewhat reductionist in vitro model, the authors suggest a protective feedback mechanism between type 2-secreting uILC2 and IL-1b-expressing DCs. This is an interesting concept that still needs a formal confirmation in vivo. Are uILC2 also subjected to plasticity upon IL-1b treatment (Ohne et al. 2016)?

In conclusion, the authors provide a well-conceived study that will be useful for reproductive and tissue immunologists. The data are collected using validated models and methodology and analyzed in a solid manner and can be used as a starting point for further mechanistic studies, assessing the protective potential of uILC2 in pregnancy during infections.

Reviewer #3 (Public Review):

The authors show that ILC2s seem to be important during pregnancy to achieve an optimal fetal growth. This is an important finding to the field and provides ILC2s with new roles distinct from parasite protection and allergy inducers. However, the fetal weight restriction phenotype does not seem very striking. Moreover, the mechanism by which ILC2s promote homeostasis in pregnancy are not well shown. The data shown in Figure 3 is overall a bit confusing and does not lead the reader to the conclusions stated in the text. Figure 4 conclusions are not very informative. The authors also show that ILC2s are protective to fetal loss during LPS infection. Again, the means by which ILC2s could be doing so are not well presented and the supporting data not fully convincing. Throughout the manuscript, the authors present quantitative data as fold change, expressing data in fold change is not as clear as showing actual numbers of cells for each group. Moreover, they should show the flow cytometry plots and gating strategy for all their FACS analysis.