The cIAP ubiquitin ligases sustain type 3 γδ T cells and ILC during aging to promote barrier immunity

The neonatal period is the time when our immune system is imprinted with life-long functional characteristics that maintain immunity to infection and prevent autoimmune pathology. Microbial colonization and developmentally regulated transcriptional programs cooperate to shape innate and adaptive lymphocytes into distinct specialized lineages that co-exist in equilibrium and respond ad hoc (Eberl, 2016). Failure to convey these environmental and molecular cues during neonatal life often results in irreversible dysfunction later on. Hence, early dysbiosis impairs type 3 immunity and potentiates susceptibility to type 2–driven allergy (Cahenzli et al., 2013). Similarly, blockade of key signaling pathways during neonatal life can permanently change cellular niches (Kadekar et al., 2020). Therefore, elucidating the molecular signatures that operate early in life is of great importance for understanding how immunity develops.

Mouse γδ T cells present a well-established example of an immune population that is heavily dependent on an unperturbed neonatal period. In this regard, intestinal intraepithelial (IE) γδ T cells develop during neonatal and prepubescent life through butyrophylin-driven interactions with the epithelia (Di Marco Barros et al., 2016). This provides a necessary defense mechanism against infection within the IE lymphocyte compartment (Hoytema van Konijnenburg et al., 2017). Lamina propria (LP) IL-17–producing γδ T cells establish mixed type 3 and type 1 transcriptional programs within the first week of life through the transcription factor STAT5 (Kadekar et al., 2020). Thus, early-life establishment of the γδT17 compartment is critical to protect from neonatal and adult infections (Chen et al., 2020; Sheridan et al., 2013). In a similar manner, impaired microbial colonization of the ocular or oral mucosa results in drastically altered IL-17–producing γδ T (γδT17) cell numbers in the conjunctiva (St. Leger et al., 2017) and cervical lymph nodes (LNs; Fleming et al., 2017). Again, paucity in such γδT17 cell populations is associated with impaired antimicrobial responses in the eye and oral cavity (Conti et al., 2014; St. Leger et al., 2017) and resistance to pathogenic inflammation (Cai et al., 2011; Sandrock et al., 2018; McGinley et al., 2020).

The innate lymphoid cell (ILC) compartment is also dependent on early life events, while their function during the neonatal period is critical for the establishment of the intestinal immune system (Spits et al., 2013). In this regard, although dysbiosis does not affect ILC development, it results in altered transcriptional and epigenetic profiles of all ILC subsets (Gury-BenAri et al., 2016). Similar to LP γδT17 cells, group 3 ILC (ILC3) acquire expression of the transcription factor Tbet and the type 1 cytokine IFN-γ during neonatal life, which allows them to clear intracellular bacterial infections (Klose et al., 2013). Moreover, group 2 ILC (ILC2) undergo an IL-33–dependent maturation step in the neonatal lung, allowing their cytokine responsiveness in adult mice (Steer et al., 2020). Importantly, ILC3 induce the maturation of intestinal cryptopatches into isolated lymphoid follicles (ILFs) during the first 3–4 wk of life (Kiss et al., 2011; Kruglov et al., 2013). Evidently, perturbations of γδT17 and ILC3 development during the early stages of life will have a substantial impact on the quality of immunity while aging. The molecular pathways that control the transition of these cells from neonatal life to adolescence and adulthood are poorly understood.

The E3 ubiquitin ligases cellular inhibitor of apoptosis protein 1 and 2 (cIAP1/2) catalyze both degradative lysine(K)48 and stabilizing K63 ubiquitination and act as the main molecular switches for the activation of the canonical and non-canonical NF-κB pathway (Silke and Meier, 2013). The presence of cIAP1/2 downstream of TNF receptor 1 (TNFR1) determines whether a cell will initiate the canonical NF-κB pathway or die by apoptosis or necroptosis in response to TNF (Annibaldi and Meier, 2018). They achieve this by ubiquitinating receptor-interacting kinase-1 (RIPK1; Silke and Meier, 2013). However, cIAP1/2 are mostly recognized as negative regulators of the non-canonical NF-κB pathway. Hence, in all cell types, cIAP1/2 associate in a heterocomplex with TNF receptor–associated factor (TRAF) 2, TRAF3, and NF-κB–inducing kinase (NIK), whereby they induce K-48 ubiquitination of NIK, resulting in its continuous proteasomal degradation (Zarnegar et al., 2008; Varfolomeev et al., 2007; Vince et al., 2007). Breakdown of the TRAF2-TRAF3-cIAP1/2-NIK complex either following ligation of TNF superfamily receptors that recruit TRAF2-TRAF3 in their intracellular domain (e.g., TNFR2, LTβR, and CD40) or by cIAP1/2 depletion, liberates NIK, which initiates the cascade necessary for nuclear translocation of the non-canonical NF-κB transcription factors RelB and p52 (Vallabhapurapu et al., 2008; Matsuzawa et al., 2008).

In the present study, we demonstrate a necessary role for cIAP1/2 in sustaining γδT17 cells and ILC3 at the late neonatal and prepubescent stages of life, thus impacting the magnitude of inflammatory and antibacterial immune responses. Deficiency in cIAP1/2 began to have an impact only during late neonatal life by reducing expression of the lineage-defining transcription factors cMAF and RORγt, which was followed by an apparent block in cytokine-induced proliferation. When animals entered prepubescence and early adolescence, cIAP1/2 deficiency resulted in the progressive loss of γδT17 cells. This was independent of TNFR1-induced canonical NF-κB or cell death. In contrast, cIAP1/2-deficient prepubescent γδT17 cells displayed enhanced nuclear translocation of RelB, which demonstrates evidence of overt activation of the non-canonical NF-κB pathway. Intestinal ILC3 also relied on intact cIAP1/2 during the same time period, with their numbers being drastically reduced in adulthood. Paucity in ILC3 coincided with ILF involution. Finally, mice with targeted deletion of cIAP1/2 in γδT17 cells and ILC3 responded suboptimally to cutaneous inflammatory challenge and failed to control intestinal bacterial infection.

Using acute, SMAC (second mitochondria-derived activator of caspases) mimetic–driven antagonization and in vitro techniques, we showed before that cIAP1/2 are important for T_H17 differentiation through modulation of the non-canonical NF-κB pathway (Rizk et al., 2019). In order to understand the in vivo importance of cIAP1 and cIAP2 in RORγt-expressing immune cells, we crossed Rorc-Cre (RORγt^CRE) mice (Eberl and Litman, 2004) with mice that were floxed for Birc2 (cIAP1^F/F) and knocked out for Birc3 (cIAP2^−/−; Gardam et al., 2011). This generated mice with RORγt-driven deletion of cIAP1 (referred to thereafter as ΔIAP1) and generalized deletion of cIAP2 (referred to thereafter as ΔIAP2), as well as the corresponding Cre-negative littermate controls (WT; Fig. 1 A). ΔIAP1 and ΔIAP1/2 mice were viable, produced offspring at expected rates, and did not develop any observable spontaneous disease phenotypes. They contained a full set of LNs (inguinal, brachial, axillary, and mesenteric) and Peyer’s patches, indicating unperturbed lymphoid tissue development, while total numbers of CD4⁺ T and B cells were normal but γδ T cell numbers, especially CD27⁺ cells, were elevated (Fig. S1 A). In line with this, IFN-γ–producing γδ T cells were also increased (Fig. S1 B).

We next analyzed some of the major IL-17–producing populations in the LN and small intestinal and colonic LP (siLP and cLP, respectively). Compared with littermate controls, ΔIAP1/2 mice produced slightly elevated levels of IL-17A within the CD4⁺TCRβ⁺ compartment in the LN (pool of inguinal, brachial, and axillary) but not the gut (Fig. S1 C), suggesting that in these animals, steady-state production of IL-17A by CD4⁺ T cells is not defective. Staining for IL-22 following overnight stimulation with IL-23 yielded the same answer (Fig. S1 C). However, there was a marked reduction in γδ-associated IL-17A and IL-22 production in LN (Fig. 1 B) and IL-17A in the gut (Fig. 1 C). This was accompanied by a dramatic loss in LN TCRγδ⁺CD27⁻CD44^hiCCR6⁺, which are the IL-17–producing γδ T cells (Ribot et al., 2009; Haas et al., 2009; Fig. 1 D and Fig. S1 D), and gut Tbet⁺RORγt⁺ (Kadekar et al., 2020; Fig. 1 E) γδT17 cell numbers. Although cIAP1 and cIAP2 individually did not contribute to this phenotype in the LN, ΔIAP2 mice had significantly reduced Tbet⁺RORγt⁺ γδ T cell numbers in the gut (Fig. 1 E). We additionally found significantly reduced γδT17 cells in the lungs of ΔIAP1/2 mice (Fig. S1 E). Similar to γδT17 cells, there were significantly reduced non-CD4 IL-17–producing lymphocytes in the LNs of ΔIAP1/2 mice (Fig. S1 F). Interestingly, ΔIAP1/2 mice had reduced numbers of CD4⁺CD8αα⁺ IE cells (Fig. S1 G).

In the skin, CD3^loVγ5⁻TCRγδ⁺CCR6⁺ cells, which represent the γδT17 population (Haas et al., 2009, 2012), were also reduced significantly in the absence of cIAP1 and cIAP2 (Fig. 2 A). When we analyzed the two major γδT17 subpopulations (Vγ4⁺ versus Vγ4⁻; Vγ nomenclature by Heilig and Tonegawa; Heilig and Tonegawa, 1986), we found that in the skin, cIAP1, but not cIAP2, was required for Vγ4⁻ cells, whereas the Vγ4-expressing population was only affected by the absence of both cIAP1 and cIAP2 (Fig. 2 B). In the LN, we did not observe differential regulation of either Vγ4⁺ or Vγ4⁻ cells (Fig. S1 H). Collectively, these data suggest that cIAP1/2 are important for the development and/or homeostatic maintenance of γδT17 cells. Our findings additionally pinpoint a differential and non-redundant role of cIAP1 and cIAP2 in these cells that is organ and subset specific. In this regard, whereas skin γδT17 cells depended more on cIAP1, gut γδT17 cells depended more on cIAP2.

Next, we investigated whether the defect we observed in ΔIAP1/2 mice was cell intrinsic. To this end, we set up mixed bone marrow (BM) chimeras where WT CD45.1⁺CD45.2⁺ hosts were sublethally irradiated and reconstituted with a mixture of 1:1 CD45.1⁺ WT and CD45.2⁺ ΔIAP1/2 BM cells (Fig. 3 A). We found that, under these conditions, WT LN γδT17 cells outcompeted their ΔIAP1/2 counterparts (Fig. 3 B), indicating that the phenotype we observed in intact mice was cell intrinsic. Interestingly, CD27⁺ γδ T cells derived from ΔIAP1/2 BM were slightly less competitive than WT (Fig. 3 B). In contrast, both CD3⁻ populations and B cells from ΔIAP1/2 BM were more competitive than their WT counterparts (Fig. 3 C). This indicated that the reduced competitiveness of γδT17 and CD27⁺ γδ T cells was not due to defective ΔIAP1/2 BM reconstitution. We could not reconstitute gut RORγt⁺Tbet⁺ γδT17 cells irrespective of the BM source (Fig. 3 D), suggesting that this population requires either thymus-originated γδ T cells or a neonatal microenvironment to develop fully. In contrast, lack of cIAP1 and cIAP2 did not impinge on the reconstitution of gut Tbet⁺RORγt⁻ γδ T cells (Fig. 3 E). Likewise, we could only recover WT LN γδT17 cells when we reconstituted ΔIAP1/2 hosts with WT or a 1:1 mix of WT and ΔIAP1/2 BM (Fig. 3, F–H), while CD27⁺ γδ T cells from WT or ΔIAP1/2 BM cells were equally competitive (Fig. 3 H). As before, we could not reconstitute RORγt⁺Tbet⁺ γδT17 cells in the gut (Fig. 3 I).

As γδT17 cells develop perinatally in the thymus and undergo a rapid neonatal reprogramming within the tissues they localize at, we reasoned that if generated from BM stem cells, they might have different developmental or homeostatic requirements for cIAP1 and cIAP2. To address this issue, we purified γδ T cells from the thymi of 1-d-old WT or ΔIAP1/2 mice and transferred them to RAG1^−/− recipients (Fig. 3 J). We found that 12 wk after transfer, the γδT17 cell compartment was reconstituted in the LN; however, we recovered significantly more WT than ΔIAP1/2 cells (Fig. 3 K). As with the BM chimeras, we could not reconstitute intestinal RORγt⁺Tbet⁺ γδT17 cells, suggesting that this population requires a neonatal microenvironment (Fig. 3 L). Reconstitution of CD27⁺ γδ T cells was independent of cIAP1 and cIAP2 (Fig. 3 K). Taken together, our data show that γδT17 cells require cIAP1 and cIAP2 intrinsically.

In addition to preventing spontaneous activation of the non-canonical NF-κB pathway, cIAP1/2 are necessary to convey the canonical NF-κB downstream of TNFR1 whereas, in their absence, TNF–TNFR1 interactions can lead to RIPK1-mediated cell death via apoptosis or necroptosis (Annibaldi and Meier, 2018). Mice deficient in TNFR1 had an intact γδT17 cell population (Fig. S2 A), suggesting that the canonical NF-κB pathway downstream of TNFR1 is not responsible for the phenotype of ΔIAP1/2 mice. Since TNF is highly upregulated during the weaning reaction (Al Nabhani et al., 2019), we next investigated whether TNF-induced cell death played a role. To achieve this, we initially analyzed mice that were deficient in cIAP2 and expressed a ubiquitin-associated (UBA) domain mutant form of cIAP1 unable to K48 ubiquitylate and suppress RIPK1 (Annibaldi et al., 2018). Thus, these mice are more sensitive to TNF-induced cell death (Annibaldi et al., 2018). In UBA-mutant mice, γδT17 cells were not affected (Fig. S2 B), suggesting that these cells are not susceptible to death by homeostatic levels of TNF. In order to test this directly in ΔIAP1/2 mice, we began injecting 1-wk-old neonates with neutralizing anti-TNF antibody until the animals were 12 wk old (Fig. S2 C). We could not rescue the γδT17 population in either gut or LNs (Fig. S2, D and E), indicating that TNF-induced death is unlikely to play a major role in regulating these cells in the absence of cIAP1/2. Therefore, TNF–TNFR1 interactions are not responsible for the ΔIAP1/2 phenotype, suggesting that overt activation of the non-canonical NF-κB pathway could play a key role.

To assess the impact of cIAP1/2 on embryonic γδT17 cell development, we enumerated thymic cell numbers in newborn ΔIAP1/2 mice and found them similar to littermate controls (Fig. 4 A), despite the efficient deletion of Birc2 (cIAP1; Fig. S3 A). Production of IL-17A/F was unchanged at this stage (Fig. S3 B). This suggested that the major impact of cIAP1/2 occurs post-embryonically. We therefore tracked LN γδT17 cells, defined phenotypically as CD27⁻CD44^hi, during neonatal, post-neonatal (average weaning time at 3 wk), and adult life (mating age of 8 wk). We did not find any differences in cell numbers until week 5 of age (Fig. 4 B). This suggested that cIAP1 and cIAP2 are only required to sustain γδT17 numbers following weaning. After week 5, ΔIAP1/2 γδT17 cells failed to expand and began to decline progressively during aging (Fig. 4 B). Interestingly, cIAP1/2 deficiency appeared to affect earlier or to a greater extent the Vγ6⁺ γδT17 population (Fig. 4 C). To confirm that the cells are missing from adult life and have not converted to a non-γδT17 population, we crossed ΔIAP1/2 with the ROSA26-LSL-RFP strain, so that RFP permanently marks all current and “ex” RORγt-expressing cells. We found no evidence of γδT17 conversion to other populations, either in adult or 4-wk-old mice (Fig. 4 D). These data suggested that loss of cIAP1/2 is unlikely to result in the conversion of γδT17 cells into non-γδT17 populations.

We next investigated potential mechanisms by which cIAP1/2 were promoting the maintenance of γδT17 cells. The inability of the cells to increase in numbers during aging raised the hypothesis that cIAP1/2 may be regulating responsiveness to cytokines that induce proliferation. We thus isolated 4-wk-old LN cells and treated them in vitro with IL-7 or a combination of IL-1β+IL-23. We found that ΔIAP1/2 γδT17 cells were slower in entering the cell cycle with most cells arrested in G0 (Fig. 4, E and F). This defect in proliferation was also observed following stimulation with either IL-2 or TCR crosslinking (Fig. S3 C). Therefore, cIAP1/2 are important for γδT17 cell cycle progression.

STAT5 is important for the proliferation and development of neonatal γδT17 cells (Kadekar et al., 2020) while STAT3 is critical for IL-23 signaling but not γδT17 development (Agerholm et al., 2019). We therefore tested whether lack of cIAP1/2 affected phosphorylated and total levels of STAT5 and STAT3. We found that STAT5 or pSTAT5 levels were unchanged (Fig. S3, D and E). pSTAT3 levels in response to IL-23 were slightly reduced (Fig. S3 D); however, total STAT3 was not changed (Fig. S3 E). Hence, the defects in the proliferation of ΔIAP1/2 γδT17 cells are unlikely to be driven by STAT5 or STAT3.

We have shown before that ablation of cIAP1/2 in T cells downregulates cMAF, a lineage-determining transcription factor for γδT17 cells (Zuberbuehler et al., 2019), in a NIK- and RelB-dependent mechanism (Rizk et al., 2019). We thus hypothesized that lack of cIAP1/2 may influence the expression of cMAF. In the newborn thymus, expression of cMAF as well as RORγt was unchanged (Fig. S3 F). At week 1 after birth, we observed a slight reduction in the expression of RORγt and cMAF (Fig. S3 G). However, at 3 wk of age, expression of RORγt and cMAF was significantly reduced (Fig. 5 A). Furthermore, we observed a modest but significant reduction in IL-17A production in 3-wk-old ΔIAP1/2 γδT17 cells (Fig. 5 B). In the intestine, RORγt⁺ γδ T cells express high levels of CD127 (IL-7Rα) and intermediate levels of CD45 (Fig. 5 C). Due to the lack of other reliable surface markers to identify these cells in the gut, we gated TCRγδ⁺CD45^intCD127⁺ cells and quantified numbers as well as expression of RORγt and cMAF. Similar to the LNs, numbers in the siLP did not change in 4-wk-old ΔIAP1/2 mice (Fig. 5 D); however, there was a significant reduction in the levels of RORγt and cMAF (Fig. 5 E).

Furthermore, we investigated the expression of RelB in newborn thymic γδ T cells and found a significant upregulation of Relb mRNA in ΔIAP1/2 γδT17 but not in CD27⁺ γδ T cells (Fig. S3 H). We additionally examined the extent of RelB nuclear translocation in γδT17 cells from 4-wk-old WT, ΔIAP2, and ΔIAP1/2 mice. We found that the levels of nuclear RelB in ΔIAP1/2 γδT17 cells were significantly higher in comparison with WT cells (Fig. 5 F), arguing for a role of the non-canonical NF-κB pathway in the downregulation of RORγt and cMAF. Therefore, cIAP1/2 are required during late neonatal life to maintain the expression of the transcription factors RORγt and cMAF and to sustain normal γδT17 numbers.

Next, we investigated whether cytokines that activate γδT17 cells could regulate the expression of RORγt and cMAF from 4-wk-old mice. Culture with IL-7 did not influence the expression of either transcription factor (Fig. 6, A–C); however, a combination of IL-1β and IL-23 resulted in partial restoration of RORγt but not cMAF in ΔIAP1/2 γδT17 cells (Fig. 6, A–C). Interestingly, IL-1β+IL-23 resulted in the downregulation of cMAF in WT cells (Fig. 6, A and B). We additionally observed that the ΔIAP1/2 γδT17 cells that acquired RORγt were the cells that entered G1 in response to IL-1β+IL-23 and to a lesser extent in response to IL-7 (Fig. 6 D). In the imiquimod (IMQ)-driven psoriasiform dermatitis model, IL-23, IL-1β, and IL-7 drive γδT17 cell expansion as well as production of IL-17 and IL-22 in the LN and skin (Michel et al., 2012; Cai et al., 2011, 2014). We thus treated WT, ΔIAP1, ΔIAP2, and ΔIAP1/2 4-wk-old mice with IMQ for 7 d and assessed the expression of RORγt and cMAF in γδT17 cells. We found that IMQ treatment partially restored expression of both RORγt and cMAF in ΔIAP1/2 γδT17 cells in the LNs (Fig. 7 A), suggesting that inflammation can rescue the ΔIAP1/2 phenotype.

We then investigated whether rescue of RORγt and cMAF was sufficient for ΔIAP1/2 γδT17 cells to mount an immune response. We observed that despite an increase in Ki67 expression (Fig. S4 A), ΔIAP1/2 γδT17 numbers did not increase in either the LNs or skin (Fig. 7 B). Evaluation of cytokine production revealed substantial regional differences between LN and skin in ΔIAP1/2 mice. Thus, whereas in the LN, ΔIAP1/2 γδT17 cells increased (albeit significantly less than their WT, ΔIAP1, and ΔIAP2 counterparts) their production of IL-17A following IMQ treatment, this was not the case in the skin (Fig. 7 C). In contrast, IL-22 production in LNs was significantly reduced while it was relatively normal in the skin (Fig. 7 D). The CD4⁺ T cell response to IMQ was not defective and slightly stronger in ΔIAP1/2 mice (Fig. S4 B). The extent of skin inflammation, as measured by epidermal thickening, was not different between ΔIAP1/2 and control mice, reflecting both the partial rescue of the γδT17 as well as the slightly exaggerated CD4⁺ T cell response (Fig. S4 C).

The data suggest that although at a young age cIAP1/2 regulate proliferation, transcriptional stability, and cytokine production, strong inflammatory stimuli can, to a certain extent, overcome this regulatory checkpoint and revive γδT17 cell responses. These results additionally indicate that the extrathymic expression and biological impact thereafter of RORγt and cMAF can be dynamic and under the control of multiple microenvironment cues.

ILC3 share many functional characteristics and transcription factor requirements with γδT17 cells, including constitutive expression of RORγt and cMAF (Zuberbuehler et al., 2019; Parker et al., 2019). We, therefore, investigated the impact of cIAP1 and cIAP2 deficiency in intestinal ILC3 populations. Similar to γδT17 cells, LP Tbet⁺ and Tbet⁻ ILC3 numbers were reduced in ΔIAP1/2 mice (Fig. 8, A–C; and Fig. S5 A). ILC3 numbers were also reduced in mesenteric LNs (Fig. S5 C). As expected, ILC2 and ILC1 numbers were not affected (Fig. S5 B). Next, we tested whether ΔIAP1/2 ILC3 converted to a RORγt⁻ population and hence performed a lineage-tracing experiment using the RORγt-RFP reporter mice described above. We found that within the ILC population, there was a 10-fold reduction in RFP⁺ cell numbers, suggesting that in the absence of cIAP1 and cIAP2, there is a loss of ILC3 rather than conversion to a non-ILC3 population (Fig. S5 D). Similar to γδT17 cells, we found that ILC3 numbers did not expand after weaning (Fig. 8 D). Next, we tested whether the loss of cIAP1/2 affected ILC3 or their progenitors during embryonic life. Hence, we enumerated LP CD4⁺ and CD4⁻ ILC3 numbers at embryonic day 18.5 (E18.5), as well as numbers of LP and fetal liver ILC progenitors (CD45⁺IL-7Rα⁺PLZF⁺PD1⁺ FLT3⁻Gata3⁻RORγt⁻; Stehle et al., 2021). We found no evidence that loss of cIAP1/2 affected ILC3 or their progenitors during embryonic life (Fig. 8 E). Next, we investigated whether the ILC3 defect in ΔIAP1/2 mice was cell intrinsic. To this end, using mixed BM chimeras, we found that WT ILC3 outcompeted their ΔIAP1/2 counterparts (Fig. 8, F and G), indicating that the phenotype we observed in intact mice was cell intrinsic. There was equal reconstitution capacity of GATA3-expressing ILC2 derived from WT or ΔIAP1/2 BM (Fig. S5 E), demonstrating the specificity of the defect within RORγt-expressing populations. We obtained similar results when we reconstituted ΔIAP1/2 hosts with a 1:1 mix of WT and ΔIAP1/2 BM (Fig. 8 H and Fig. S5 F). Further, cIAP1/2-deficient ILC3 cells were not rescued by treatment with anti-TNF (Fig. S5 G).

ILC3 are necessary for the maturation of cryptopatches to ILFs during the first weeks of life. Qualitative analysis by immunofluorescence indicated that ILFs in ΔIAP1/2 mice were defective (Fig. 8 I). Despite the lack of ILFs, the production of IgA was not defective (Fig. S5 H). Collectively, these data show that cIAP1 and cIAP2 are necessary for intestinal ILC3 to expand during the post-weaning period and to induce the formation of ILFs.

It has been demonstrated that ILC3 are important to control infection by the attaching and effacing bacterium C. rodentium (Guo et al., 2014, 2015; Bauché et al., 2020), a widely used model for human enteropathogenic Escherichia coli infections (Silberger et al., 2017). We therefore reasoned that ΔIAP1/2 mice may be defective in mounting a protective response to C. rodentium. We infected ΔIAP1/2 mice and their respective controls with 2 × 10⁹ CFU of C. rodentium through oral gavage and followed weight loss as a surrogate marker for disease. We found that by 11 d after infection, ΔIAP1/2 mice lost ∼20% of their body weight, while all other strains did not (Fig. 9 A). At this time point and due to ethical constraints, all animals were sacrificed, and we analyzed bacterial loads and the immune response in the colon. ΔIAP1/2 mice had significantly higher colonic bacterial load than controls (Fig. 9 B). This was associated with compromised IL-22 production from the ILC3 compartment (Fig. 9, C and D).

Although ILC3 are important to protect from C. rodentium infection, a T_H17 and T_H22 response is also required, as evidenced by the susceptibility of RAG1^−/− mice to this pathogen (Silberger et al., 2017). We therefore additionally analyzed the CD4⁺ T cell response in the colon. The number of total CD4⁺ T cells were not changed in infected ΔIAP1/2 mice (Fig. 9 E). However, there was a significant reduction in RORγt⁺Tbet⁻ CD4⁺ T cells (Fig. 9 E), which was accompanied by reduced levels of IL-17A (Fig. 8 F). Despite normal numbers of RORγt⁺Tbet⁺ CD4⁺ T cells (Fig. 9 E), IL-17A⁺IFN-γ⁺ cells were significantly reduced in infected ΔIAP1/2 mice (Fig. 9 F). However, production of IL-22 was not defective in the absence of cIAP1 and cIAP2 (Fig. 9 F). Collectively, our data suggest that cIAP1 and cIAP2 are required within the ILC3 and T_H17 compartments to control intestinal bacterial infection.

The murine gut is colonized with microbiota after birth and during neonatal life. At around weaning (3–4 wk of age) and while transitioning from milk to solid food, the microbiota induces an immune response, a weaning reaction, that is important for subsequent cell maturation (Al Nabhani et al., 2019). Since ΔIAP1/2 γδT17 cells and ILC3 began declining in numbers around the age of weaning, we speculated that the microbial load may contribute to the phenotype. To test this, we started treating pregnant females at E18.5 with broad-spectrum antibiotics (Abx). Treatment continued throughout the neonatal and weaning period and until the litters reached 12 wk of age (Fig. 10 A). In the LN, γδT17 cell numbers declined in response to Abx (Fig. 10 B). However, there was still a significant difference between ΔIAP1/2 and littermate control cell numbers (Fig. 10 B). The number of ΔIAP1/2 γδT17 cells were not changed irrespective of Abx treatment (Fig. 10 B). We additionally observed that the Abx partially restored RORγt but not cMAF expression in the remaining ΔIAP1/2 γδT17 cells (Fig. 10 C).

Similarly, siLP ΔIAP1/2 γδT17 numbers were reduced in Abx-treated compared with littermate control Abx-treated mice (Fig. 10 D). However, direct comparison of Abx-treated and untreated animals showed a partial but clear restoration of ΔIAP1/2 γδT17 cell numbers (Fig. 10 D). Interestingly, ΔIAP1/2 RORγt⁻Tbet⁺ γδ T cell numbers were increased following Abx treatment (Fig. 10 D). When we examined siLP ILC3, we found that their numbers were not affected by Abx treatment (Fig. 10 E). Similar to γδT17 cells, we observed a partial rescue of Abx-treated ΔIAP1/2 ILC3, although this was not statistically significant (Fig. 10 E). Hence, the presence of the microbiota exaggerated the demise of intestinal ΔIAP1/2 γδT17 and ILC3 populations, although the impact on the latter was less pronounced.

In the present study, we demonstrate that the E3 ubiquitin ligases cIAP1 and cIAP2 are necessary for γδT17 cells to transition through to prepubescent life by regulating cytokine-mediated proliferation and stable expression of the lineage-defining transcription factors cMAF and RORγt. Thus, during aging, cIAP1 and cIAP2 are required in a cell-intrinsic manner to maintain cMAF and RORγt levels and to allow cells to enter the cell cycle in response to IL-7, IL-1β, and IL-23. Consequently, γδT17-driven inflammatory responses in the skin and draining LNs of prepubescent ΔIAP1/2 mice are blunted despite normal cell numbers, while by the time animals reach adulthood, γδT17 populations are deficient in gut, skin, and LNs. Mechanistically, our data suggest that this is independent of TNF and TNFR1 and most likely through overt activation of the non-canonical NF-κB pathway. Similar to γδT17, ILC3 required cIAP1 and cIAP2 expression during the post-weaning period to expand, be maintained until adult life, and induce formation of intestinal ILFs. The ILC3 deficit in ΔIAP1/2 mice, together with a defective T_H17 response, correlated with a profound inability to control intestinal bacterial infection. Finally, our data additionally suggest that the progressive demise of γδT17 cells and ILC3 in ΔIAP1/2 mice can be partially counteracted by ablating the microbiota.

The IL-17–producing γδ T cell subset is an innate-like unconventional lymphocyte that is important in many immunological processes ranging from antimicrobial protection to pathogenic inflammation and cancer (Patil et al., 2015). γδT17 cells are preprogrammed and functionally mature in the embryonic thymus in mouse and human (Ribot et al., 2009; Haas et al., 2012). They are exported into peripheral and secondary lymphoid tissues after birth, and evidence suggests that they go through a second wave of transcriptional and functional programming during neonatal life within the tissues they occupy (Kadekar et al., 2020; Wiede et al., 2017). The molecular cues that γδT17 cells receive within the tissues during that period are obscure. Our data show that the E3 ligases cIAP1 and cIAP2 are required during late neonatal and early prepubescent life in a cell-intrinsic manner for cytokine-induced proliferation, to sustain transcriptional stability, and allow optimal inflammatory responses. Interestingly, our data further suggest that the progressive demise of γδT17 cells inflicted by loss of cIAP1/2 can be partially rescued in the gut when the microbiota is impaired. At this stage, we can only speculate that the presence of a normal microbial load may act as an additional stress factor for the already “fragile” cells that lack cIAP1/2. Collectively, this work establishes cIAP1/2 as critical molecular regulators of committed tissue-resident γδT17 cells and underpins the existence and importance of post-thymic temporal events necessary for these cells to be maintained during aging.

cIAP1/2 are central for TNFR1-induced canonical NF-κB activation and cell death (Mahoney et al., 2008) and necessary to suppress overt non-canonical NF-κB signaling (Vallabhapurapu et al., 2008; Zarnegar et al., 2008). TNFR1-induced apoptosis and necroptosis are fundamental biological processes regulating cell growth during development, homeostatic turnover, and even inflammatory diseases (Kalliolias and Ivashkiv, 2016). The two NF-κB pathways, on the other hand, are synonymous with cell survival, proliferation, and differentiation in ubiquitous cell populations (Hayden and Ghosh, 2011). In T cells, they are mostly active downstream of TNF superfamily receptors and the TCR (Oh and Ghosh, 2013). Although the role of several TNF superfamily receptors and ligands have been studied in γδ T cells (Powolny-Budnicka et al., 2011; Shibata et al., 2011; Silva-Santos et al., 2005), the importance of the signaling components of the NF-κB pathway had not been thoroughly investigated. Genetic and pharmacological perturbations of the TNFR1 signaling, combined with aberrant nuclear translocation of RelB showed that it is the cIAP1/2-mediated control of non-canonical NF-κB that is required for the maintenance of γδT17 cells. This agrees with CD27, a TNF superfamily receptor and potent activator of non-canonical NF-κB (Ramakrishnan et al., 2004), suppressing the γδT17 differentiation program (Ribot et al., 2009). Importantly, deletion of NIK, the kinase targeted by cIAP1/2 and responsible for activating the non-canonical NF-κB cascade, did not affect γδT17 cell development or homeostasis (Mair et al., 2015). This strongly suggests that it is the “brake” imposed by cIAP1/2 to avoid overactivation of non-canonical NF-κB that is critical and not its baseline activity.

There are a number of transcription factors that are important for the development of γδT17 cells (Parker and Ciofani, 2020). Ciofani and co-workers showed that cMAF acts early in embryogenesis to allow robust expression of RORγt and thus promote specification and stability of the γδT17 lineage (Zuberbuehler et al., 2019). How cMAF and RORγt expression is regulated, however, in γδT17 cells is not well-defined. Herein, we demonstrate that loss of cIAP1/2 results in the progressive downmodulation of cMAF and RORγt after birth, providing a molecular understanding of how lineage-defining transcription factors are regulated in these cells. Although the loss of cMAF and RORγt during embryonic development resulted in rapid loss of γδT17 cells or their progenitors in the thymus (Zuberbuehler et al., 2019; Shibata et al., 2011), we observed that in ΔIAP1/2 mice, cells persist for at least 2 wk without either transcription factor. Thus, it appears that during neonatal life, the impact of cMAF and RORγt in γδT17 cells is less pronounced. The exact molecular steps leading to cIAP1/2-dependent regulation of cMAF and RORγt are currently unclear. Our previous work showed that following cIAP1/2 inhibition, NIK-mediated RelB nuclear translocation suppressed expression of cMAF in T_H17 cells (Rizk et al., 2019). It is plausible, therefore, that accumulation of non-canonical NF-κB signaling directly suppresses cMAF, which subsequently suppresses RORγt. Intriguingly, cytokine stimulation and inflammation could partially restore the expression of cMAF and RORγt, indicating a certain degree of transcriptional plasticity.

In addition to γδT17 cells, cIAP1/2 were necessary for the establishment of a normal ILC3 population during the post-weaning period in the gut and the formation of ILFs, as well as protection from intestinal extracellular bacterial infection. During infection, we additionally found that cIAP1/2 were critical for the generation of IL-17⁺ and IL-17⁺IFN-γ⁺ CD4⁺ T cells, which have been associated with protection against pathogens, or tissue damage in the context of inflammation (Omenetti et al., 2019). We and others have previously reported that the cIAP–non-canonical NF-κB axis is necessary for T_H17 differentiation and successful IL-17–driven responses (Rizk et al., 2019; Kawalkowska et al., 2019), while NIK was shown to be important for the generation of neuropathogenic T_H17 cells (Lacher et al., 2018). Moreover, NIK expression and activation of the non-canonical NF-κB pathway in dendritic cells indirectly regulate the maintenance of both T_H17 cells and ILC3 (Jie et al., 2018). Our current data extend and broaden the immunological importance of this pathway. We would like to propose that through regulation of non-canonical NF-κB, cIAP1/2 are master regulators of innate and adaptive type 3 immunity. Their requirement is necessary during neonatal life to establish functional innate and innate-like type 3 immune cell populations, whereas in the adult, they support differentiation of antigen-dependent adaptive type 3 cells.

All animals were bred and maintained in-house at the Technical University of Denmark BioFacility under a breeding license approved by the Danish Animal Experiments Inspectorate. All experiments involving mice were performed with the approval of the Danish Animal Experiments Inspectorate. cIAP1^F/F and cIAP1^F/FcIAP2^−/− mice were provided by Prof. W. Wei-Lynn Wong at the University of Zurich, Zurich, Switzerland with the permission of Prof. John Silke, VIC Australia. RORγt-Cre mice were provided by Prof. Gerard Eberl at the Pasteur Institute, Paris, France. ROSA26-floxSTOPflox-RFP mice were from the Swiss Immunological Mouse Repository. The mice were bred by crossing the RORγt-Cre strain to the cIAP1^F/F or crossing RORγt-Cre to the cIAP1^F/F cIAP2^−/− strain. In this work, WT mice are RORγt-Cre⁻ cIAP1^F/F cIAP2^+/+, ΔIAP1 are RORγt-Cre⁺ cIAP1^F/F cIAP2^+/+, ΔIAP2 are RORγt-Cre⁻ cIAP1^F/F cIAP2^−/−, and ΔIAP1/2 are RORγt-Cre⁺ cIAP1^F/F cIAP2^−/−. LNs from TNFR1^−/− mice were provided by Prof. William Agace at Lund University, Lund, Sweden, while LNs from cIAP1^UBA mutant mice were provided by Prof. Pascal Meier at The Institute of Cancer Research, London, UK.

For all preparations of single-cell suspensions and cell cultures, RPMI 1460 (Invitrogen) supplemented with 10% heat-inactivated FBS (GIBCO), 20 mM Hepes pH 7.4 (Gibco), 50 µM 2-mercaptoethanol, 2 mM L-glutamine (Gibco), and 10,000 U/ml penicillin-streptomycin (Gibco) was used. Where indicated, IMDM (Invitrogen) was used instead of RPMI 1460 and supplemented as aforementioned. FACS buffer was prepared by supplementing PBS with 3% heat-inactivated FBS.

Lymphocytes were isolated from peripheral LNs (axial, brachial, and inguinal), thymus, ear skin, small intestinal, and colonic LP following the previously described protocols (Kadekar et al., 2020). Lymphocytes were isolated from cervical and auricular LNs in case of IMQ-induced psoriasis.

For staining of cytokines from lymphocytes that were isolated from peripheral LNs of untreated mice, the cells were plated at a density of 10 × 10⁶ cells/ml in 1 ml of supplemented RPMI in 12-well plates. The cells were restimulated with 50 ng/ml PMA (Sigma-Aldrich), 750 ng/ml ionomycin (Sigma-Aldrich), and BD GolgiStop (containing monensin at 1:1,000 dilution; BD) and cultured for 3.5 h at 37°C. For estimation of IL-22 production by CD4⁺ and γδ T cells from homeostatic mice, the lymphocytes were first cultured overnight with 40 ng/ml rmIL-23 (R&D), then restimulated with PMA, ionomycin, and BD GolgiStop as aforementioned. The cells were then harvested and used for flow cytometry staining.

In the case of lymphocytes that were isolated from peripheral LNs or skin in IMQ experiments, the cells were plated at a density of 5 × 10⁶ cells/ml in 1 ml of supplemented IMDM in 24-well plates. The cells were restimulated with 50 ng/ml PMA (Sigma-Aldrich), 750 ng/ml ionomycin (Sigma-Aldrich), and BD Golgiplug (containing brefeldin A at 1:1,000 dilution; BD) and cultured for 3.5 h at 37°C. The cells were then harvested and used for flow cytometry staining.

Alternatively, in lymphocytes that were isolated from mesenteric LNs or cLP in C. rodentium infection experiments, the cells were plated at a density of 5 × 10⁶ cells/ml in 1 ml of supplemented IMDM in 24-well plates. The cells were subsequently treated with 40 ng/ml rmIL-23 (R&D) for 3 h, followed by 50 ng/ml PMA (Sigma-Aldrich), 750 ng/ml ionomycin (Sigma-Aldrich), and BD Golgiplug (containing brefeldin A at 1:1,000 dilution; BD) and cultured for an additional 3.5 h at 37°C.

For cell-cycle assay experiments, lymphocytes that were isolated from peripheral LNs of mice were plated at a density of 5 × 10⁶ cells/ml in 1 ml of supplemented RMPI in 24-well plates. The cells were treated with either 20 ng/ml rmIL-7 (R&D), 10 ng/ml rmIL-1β (BioLegend) + 20 ng/ml rmIL-23 (R&D), 20 ng/ml rhIL-2 (BioLegend), or anti-CD3 (2 μg/ml; clone 145-2C11) for 48 h. The cells were subsequently harvested for flow cytometry staining.

Psoriasis was induced in mice by applying 7 mg of Aldara cream (containing 5% IMQ) to the dorsal side of each ear for 7 d. Histological sections were prepared by fixing ear tissue in 4% paraformaldehyde solution in PBS overnight and then paraffin-embedded. The paraffin-embedded sections were cut and stained by H&E.

Surface antigens, intracellular cytokines, and cell cycle assay were stained for flow cytometry as previously described (Rizk et al., 2019). For transcription factor staining, the cells were first stained for live/dead discrimination followed by surface antigen staining and subsequently fixed using Foxp3 fixation/permeabilization buffer (Thermo Fisher Scientific) for 1 h at 4°C. The cells were washed once with permeabilization buffer (Thermo Fisher Scientific) then stained with the desired antibodies in Foxp3 permeabilization buffer buffer (Thermo Fisher Scientific) for 1 h at 4°C. The cells were washed once again and resuspended in FACS buffer and analyzed using BD LSRFortessa.

For staining of phospho-antigens p-STAT3 and p-STAT5, 5 million LN cells were stained with fixed viability stain-700 for 10 min on ice and washed once at 4°C. The cells were then stained with CD44-V500 and TCRγδ-APC antibodies for 30 min on ice in FACS buffer. The cells were then stimulated in Eppendorf tubes with 20 ng/ml IL-7 or 20 ng/ml IL-23 for 15 min at 37°C in supplemented RPMI 1640. Immediately after, the cells were collected by centrifugation at 4°C and 500 g for 5 min. The cells were then resuspended in 100 μl Phosflow Fix Buffer I (BD) preheated to 37°C and incubated for 15 min at 37°C. The cells were then washed once with cold PBS and resuspended in 100 μl PermBuffer III (BD), prechilled to −20°C, and incubated on ice for 30 min. The fixed cells were then collected by centrifugation at 4°C and 800 g for 8 min. Subsequently, the cells were stained with anti-pSTAT3 or anti-pSTAT5 and anti-TCRβ-APC-Cy7 and anti-CD27-PeCy7 in FACS buffer for 30 min on ice. The cells were washed once at 4°C and 800 g for 8 min and analyzed by flow cytometry.

The following antibodies were used herein at 1:200 dilution unless otherwise indicated: Fixed viability stain-700 (FVS700, 1:1,000; BD), GR-1(RB6-8C5, AF700, 1:200), anti-IL-17A (TC11-18H10; BV786 and PE), anti-IFNγ (XMG1.2; PE-Cy7, APC, BV711, and Percp-cy5.5), anti-IL-22 (1H8PWSR; PE), anti-cMAF (symOF1; PE, eF660, or Percp-Cy5.5; 5 μl/test), anti-CD4 (GK1.5; BUV395, FITC, and BV786), anti-TCRγδ (GL3; BV421 and APC), anti-KLRG1 (2F1, BV786), anti-CD27 (LG.3A10; PE-Cy7 and BV650), anti-CCR6 (140706; Alexa Fluor 647), anti-CD44 (1M7; V500), anti-CD19 (6D5; FITC), anti-TCRβ (H57-597; APC-eflour780), anti-CD3e (145-2C11, PeCF594, PE, BUV737), anti-Tbet(4B10; PeCy7), anti- CD8 (53-6.7; FITC), anti-Vγ5 (536; FITC), anti-Vγ4 (UC3-10A6; Percp-eflour710), anti-Vγ1(2.11; BV605), anti-Vγ6 (Hatano et al., 2019); conjugated to AF647), anti-GATA3 (TWAJ; Percp-eFlour710; 1:30), anti-CD45 (30-F11; PE and V500), anti-CD127 (SB/199; BUV737), CD8β (YTS156.7.7; PeCy7), anti-Thy1.2 (53-2.1, FITC, 1:200), anti-RORγt (B2D; APC, PE, and PEcf594), PLZF (R17-809; PE-eFlour610, 1:100), FLT3 (A2F10, PE, 1:100), α4β7 (DATK32; BV421, 1:100), PD-1 (J43; PEcy7,1:100), STAT3 (4G4B45; PE, 1:100), STAT5 (JJ08-78, 1:50), donkey anti-rabbit IgG (Poly4064; BV421, 1:100) p-STAT3 (4/P-STAT3; PE, 15 μl/test), p-STAT5 (47/Stat5; BV421, 5 μl/test)

For neutralization of TNF, 1-wk-old pups were weighed and i.p. injected with the anti-TNF (adalimumab, brand name HUMIRA) at 5 mg/kg body weight once a week until weaning. After weaning the mice were i.p. injected with 10 mg/kg body weight twice a week until euthanasia at ∼12 wk of age.

First, thymi from 1–2-d-old mice were isolated and crushed individually against a 70-μm filter to prepare single-cell solutions. Subsequently, total γδ T cells were enriched by magnetic depletion of CD4⁺, CD8⁺, TCRβ⁺ cells as follows: total thymocytes were resuspended in MACS buffer at 10⁸ cells/ml containing 50 μl/ml normal rat serum and 1:200 biotin-labeled anti-CD4 (GK1.5), anti-CD8 (53-6.7), and anti-TCRβ (H57-597) antibodies; the cells were incubated for 10 min at room temperature and then incubated with 75 μl/ml EasySep RaphidSphere streptavidin beads (#50001) for 2.5 min then transferred to EasySep magnet for 2.5 min. The non-bound fraction was collected by decantation and centrifugated for 5 min at 400 g at 4°C. The enriched γδ T cells from each donor mouse were resuspended in PBS and then i.v. injected into the tail vein of a RAG1^−/− host. The RAG1^−/− hosts were euthanized for collection of organs after 12 wk.

The BM cells for reconstitution were isolated by flushing the tibia and femur, which were dissected from donor mice, with culture media. Total BM cells were then centrifuged at 400 g for 5 min at 4°C. The cells were then resuspended and passed through a 70-μm filter. Subsequently, RBCs were then lysed using RBC lysis buffer (BioLegend) and a single-cell suspension of BM cells was then prepared by passing the cells through a 40-μm filter. The prepared cells were then counted and mixed as appropriate.

Conversely, host mice were sublethally irradiated by two doses of 4.5 Gy that were at least 4 h apart. After 24 h, the hosts were reconstituted with 10 × 10⁶ BM cells that were i.v. injected into the tail vein of the host mice. The hosts were euthanized for organs after at least 12 wk.

To assess the presence of solitary intestinal lymphoid tissue in the intestines of WT or ΔIAP1/2 by confocal laser microscopy, the distal ileum was taken and flushed once with HBSS (Thermo Fisher Scientific) to remove intestinal contents. Cleaned intestines were fixed for 8 h in 4% paraformaldehyde (Sigma-Aldrich) in PBS and stored in washing buffer (PBS + 5% FCS + 0.2% Triton X-100 [Sigma-Aldrich] + 0.01% thimerosal [Sigma-Aldrich]) until further use. To prepare the collected intestines for staining, tissues were embedded in 4% UltraPure Low Melting Point Agarose (Thermo Fisher Scientific) in PBS, sectioned with a swinging blade microtome (Leica VT1200S) into 50-μm sections, and permeabilized overnight using the Foxp3 Transcription Factor Staining Buffer Set (Thermo Fisher Scientific). Permeabilized sections were stained in the supplied permeabilization buffer (Thermo Fisher Scientific) with an antibody against RORγt (AFKJS-9; unconjugated), followed by a washing step in permeabilization buffer (Thermo Fisher Scientific) and incubation with a biotinylated secondary antibody against the primary anti-RORγt antibody (biotinylated anti-rat; Jackson ImmunoResearch). To detect RORγt⁺ ILC and B cells, sections were washed again in PermBuffer and incubated in permeabilization buffer (Thermo Fisher Scientific) with antibody against B220 (RA3-6B2; AF647) and streptavidin-conjugated AF555 (Thermo Fisher Scientific), as well as DAPI (Thermo Fisher Scientific) to stain all nucleated cells. Sections were washed one more time, mounted on glass slides with ProLong Gold (Thermo Fisher Scientific), and analyzed using an LSM710 confocal laser microscope (Carl Zeiss). Images of more than five different sections per mouse were acquired with the Zeiss Zen v2.3 software (Carl Zeiss) and analyzed using Imaris v8 (Bitplane/Oxford Instruments) and Fiji v2.1.0/1.53c (Schindelin et al., 2012).

Starter cultures of C. rodentium strain DBS100 (ATCC 51459; American Type Culture Collection) were grown overnight at 37°C in Luria-Bertani (LB) medium. The cultures were then used at 5% vol/vol to inoculate a sterile LB medium. The cultures were grown at 37°C to an OD600 of 0.8–1, and the CFU count was determined from the OD600 measurement using the following formula: CFU/ml = (5 × 10⁸) × (OD) – (3 × 10⁷). Subsequently, the bacteria were collected by centrifugation at 4,000 g for 10 min. The bacterial pellet was then resuspended in LB medium to give at 2 × 10⁹ CFU/100 μl. To infect adult mice, the mice were orally gavaged with either 100 μl of C. rodentium or LB control. The mice were weighed before oral gavage and once daily until the termination of the experiment. At day 12 after infection, all mice were euthanized and dissected to collect organs and fecal samples.

The collected fecal samples were weighed and dissolved in PBS and then serially diluted. The serial dilutions were plated on Brilliance E. coli/coliform Agar (CM0956; Thermo Fisher Scientific) and incubated overnight at 37°C. C. rodentium colonies were identified as being pink colonies and enumerated, while E. coli colonies were identified as purple colonies. CFU/g stool was then calculated as previously described (Bouladoux et al., 2017).

Total lymphocytes were isolated from peripheral (axial, brachial, and inguinal), cervical, and auricular LNs of 4-wk-old mice as described above. Subsequently, total γδ T cells were enriched by magnetic depletion of CD4⁺, CD8⁺, CD19⁺, and TCRβ⁺ cells as aforementioned. The cells were then stained with FVS700 for discrimination of live and dead cells and then stained for surface antigens with the following antibodies: anti-TCRγδ (GL3; APC), anti-CD27 (LG.3A10; PE-Cy7), and anti-TCRβ (H57-597; APC-eflour780) all at 1:200 dilution. The cells were subsequently sorted in TCRγδ⁺CD27⁺ or TCRγδ⁺CD27⁻ using BD ARIA-FUSION cell sorter. The sorted cells were collected in a cell culture medium and centrifuged at 400 g for 5 min at 4°C. The cells were then fixed using Foxp3 fixation/permeabilization buffer (Thermo Fisher Scientific) for 1 h at 4°C. The cells were washed once with then stained with Foxp3 permeabilization buffer (Thermo Fisher Scientific) buffer containing anti-TCRγδ (GL3; 1:50; APC), anti-CD3e (145-2C11 or 17A2; biotin, 1:100), and anti-RelB (D-4, 1:40; Santa Cruz) for 1 h at 4°C in Foxp3 permeabilization buffer (Thermo Fisher Scientific). Again, the cells were washed once and stained with streptavidin-conjugated AF488 (1:100; BioLegend) and anti-mouse AF555 (1:100) for 1 h 4°C in Foxp3 permeabilization buffer (Thermo Fisher Scientific). The cells were then washed once more as previously and stained with DAPI to highlight cell nuclei and washed once more with PBS. Washed cells were mounted on a glass slide using ProLong Gold and imaged with an LSM710 confocal laser microscope and were acquired and analyzed with the Zen v2.3 software and Fiji v2.1.0/1.53c (Schindelin et al., 2012).

At the indicated timepoints, γδ T cells were sorted from the thymus or LNs of mice directly into RLT buffer (Qiagen) mixed with β-mercaptoethanol. RNA was then extracted using Qiagen RNAeasy Microkit following the manufacturer’s instructions. Subsequently, cDNA was prepared using Bio-Rad Iscript cDNA synthesis kit using the manufacturer’s protocol. Gene expression was then measured by RT-qPCR reactions using BioRad SSOFast EvaGreen supermix, which were run on CFX96 (Bio-Rad) and analyzed using Bio-Rad CFX manager software. The following primers were used for RT-qPCR: Actb, Fwd-5′-GGCTGTATTCCCCTCCATCG-3′, Rev- 5′-CCA​GTT​GGT​AAC​AAT​GCC​ATG​T-3′; RelB, Fwd- 5′-GCT​GGG​AAT​TGA​CCC​CTA​CA-3′, Rev- 5′-CAT​GTC​GAC​CTC​CTG​ATG​GTT-3′; Birc2, Fwd- 5′-TGC​CTG​TGG​TGG​GAA​ACT​GA-3′, Rev- 5′-GCT​CGG​GTG​AAC​AGG​AAC​A-3′.

A mix of broad-spectrum Abx consisting of 5 mg/ml streptomycin, 0.5 mg/ml vancomycin, 1 mg/ml ampicillin, and 1 mg/ml colistin was dissolved in the drinking water, which was supplemented with 2% glucose, and fed to pregnant female mice at E18.5. The water bottles containing the Abx mix were changed every 2 or 3 d and the treatment with Abx-containing water was continued until the litters were 12 wk old.

Flow cytometry data were analyzed using FlowJo version v10.8 (BD Biosciences), and all the data were statistically analyzed using Graphpad Prism v9.

Fig. S1 shows further analysis of the composition of the immune compartment in different organs of WT, ΔIAP1, ΔIAP2, and ΔIAP1/2 mice at homeostasis. Fig. S2 shows an analysis of γδT17 cells in the LNs of WT and TNFR1 KO mice and the effect of anti-TNF treatment on γδT17 cells in ΔIAP2 and ΔIAP1/2 mice. Fig. S3 shows the expression of Birc2 and Relb and the production of IL-17A and IL17F in thymic γδT17 from 1-d-old WT, ΔIAP1, ΔIAP2, and ΔIAP1/2 mice. It shows cell cycle analysis of γδT17 cells following stimulation with IL-2 or anti-CD3e. It also shows the levels of STAT3, STAT5, and p-STAT3 or pSTAT5 following cytokine stimulation ex vivo. It shows the expression of RORγt and cMAF in γδT17 cells from day 1 thymus or week 1 LNs. Fig. S4. shows an analysis of Ki67 expression in LN γδT17 cells, as well as skin and LN CD4⁺ T cell responses in 4-wk-old mice following IMQ treatment. Fig. S5 shows an analysis of ILC3s in the colon of adult WT, ΔIAP1, ΔIAP2, and ΔIAP1/2 mice at homeostasis and fate-map tracing of ILCs in the siLP. It also shows the composition of the ILC2 compartment in BM chimera experiments. It shows an analysis of ILC3 cells in the siLP following anti-TNF treatment in ΔIAP2 and ΔIAP1/2 mice. It also shows an analysis of small intestinal IgA⁺CD138⁺ plasma cells in adult WT, ΔIAP1, ΔIAP2, and ΔIAP1/2 mice at homeostasis.

The data underlying all figures are available in the published article and its online supplemental material.

We thank Prof. W. Wei-Lynn Wong (University of Zurich, Zurich, Switzerland) and Prof. John Silke (The Walter and Eliza Hall Institute, Parkville, Australia) for providing the cIAP1^F/F and cIAP1^F/FcIAP2^−/− mice. We thank Prof. Gerard Eberl (Pasteur Institute, Paris, France) for providing the RORγt-Cre mice, Prof. Pascal Meier and Dr. Alessandro Annibaldi (The Institute of Cancer Research, London, UK) for providing organs from cIAP1^UBA mutant mice, and Prof. William W. Agace (Lund University, Lund, Sweden) for providing organs from TNFR1^−/− mice. The graphical abstract was made using BioRender content (licenses: EZ25HL609E and DK25HL57LK).

Work in V. Bekiaris’s lab was supported by Lundbeckfonden grants (R366-2021-104 and R163-2013-15201) and an LEO Foundation grant (LF-OC-20-000550). J. Rizk and R. Agerholm were supported by PhD fellowships from the Technical University of Denmark. J. Rizk was supported by a Lundbeckfonden grant (R347-2020-2400).

Author contributions: J. Rizk designed and performed the experiments, analyzed the data, and wrote the manuscript. U.M. Mörbe performed the fluorescence microscopy experiments and helped analyze the data. R. Agerholm, M.V. Baglioni, E. Catafal Tardos, M.G. Fares da Silva, I. Ulmert, D. Kadekar, and M.T. Viñals performed experiments, acquired flow cytometry data, and helped analyze the data. V. Bekiaris conceptualized the study, designed and performed experiments, analyzed data, supervised the work, acquired funding, and wrote the manuscript with J. Rizk.