Flt3 Ligand Regulates the Development of Innate Lymphoid Cells in Fetal and Adult Mice

Innate lymphoid cells (ILCs) are a family of immune cells that participate in the early response to infections at mucosal surfaces (1, 2). In addition to their abundance in the gastrointestinal tract, skin, and lung (2–5), ILCs are also found in lymphoid organs such as spleen, lymph nodes (LNs), Peyer’s patches (PPs), and tonsils (6, 7). Recently, ILCs were categorized into three groups due to their transcriptional regulation of development and cytokine production (8). Group 1 ILCs consist of NK cells and ILC1s. They are characterized by the expression of NK1.1 and NKp46 (2, 9), require the transcription factor T-bet (10, 11), and produce IFN-γ (12, 13). Group 2 ILCs (ILC2s) are identified by the expression of Sca1, CD25, and CD127 (14). They depend on the transcription factors Gata3 (15) and retinoic acid–related orphan receptor (ROR) α (14, 16) and produce Th cell type 2 cytokines like IL-5 and IL-13 (14, 17). Group 3 ILCs (ILC3s) include lymphoid tissue inducer (LTi) cells that are important for the development of LNs and PPs (7, 18–20) and adult ILC3s. The latter are divided into several subsets according to their expression of CD4 and NKp46 (21, 22). They depend on the expression of RORγt (20) and produce IL-17 and/or IL-22 (23–25).

Cytokines play an important role during the development of hematopoietic cells, by supporting proliferation, survival, or lineage commitment (26). Only few cytokines are described that are required for the development and maintenance of ILCs under steady-state conditions. NK cells depend on IL-15 but not on IL-7 (27), which instead is crucial for ILC3s (13, 28–30) and ILC2s (15, 17). In addition, we have shown that stem cell factor (SCF) (29) and thymic stromal lymphopoietin (31) are important for LTi cell development. During fetal development, LTi cells arise from CD127⁺ fetal liver (FL) progenitors (32, 33). In adults, ILCs develop from common lymphoid progenitors (CLPs) in the bone marrow (BM) (9, 16, 34). Fetal and adult ILC progenitors express receptors for cytokines that are important for ILC development, like the receptors for IL-7 (CD127) and SCF (CD117) (32, 33, 35). Additionally, they express the cytokine receptor flt3 (CD135) (34, 36, 37). Evidence that flt3 ligand (Flt3L) is important for ILC development came from a study by Yang et al. (38), who reported a diminished development of lung ILC2s after transfer of BM cells from flt3^−/− mice into irradiated wild-type (WT) mice. Another study, however, reported normal ILC frequencies in the small intestine (SI) of flt3l^−/− mice (39). In order to clarify the effect of Flt3L on ILC development, we analyzed flt3l^−/− and flt3l-tg mice as well as mice treated with rFlt3L. In addition, we compared fetal and adult ILC development in flt3l^−/− and Il7^−/− mice. We show that Flt3L controls intestinal LTi cell numbers in neonatal mice and ILC numbers in the SI, mesenteric LN (mLN), spleen, and lung of adult mice by regulating the ILC progenitor pool in FL and BM. In addition, our data indicate that Flt3L acts on early ILC progenitors, whereas IL-7 is essential for later stages of ILC development.

WT C57BL/6 mice were obtained from Janvier. Rag2^−/− mice were obtained from The Jackson Laboratory. Rag2^−/−Il2rg^−/− (31), Il7^−/− (40), flt3l^−/− (41), flt3l-tg (42), CD11c-Cre (43), and R-DTA (44) mice were on a C57BL/6 background. All mice were bred and maintained under specific pathogen-free conditions according to the guidelines of the cantonal veterinarian office of Basel or the animal care and use committees of Lower Franconia, Germany. For detection of LNs, 1% Chicago sky blue 6B ink (Sigma-Aldrich) was injected s.c. into the footpads of mice 2 d before analysis. For each strain, the total number of LNs of all animals was counted for each location and divided by the number of animals analyzed. The frequency of LNs (in percentage) per location in mutant strains was calculated relative to 100% LNs in WT mice. For Flt3L treatment, mice were injected i.p. for 10 d with 20 μg recombinant Flt3L (rFlt3L)/d and sacrificed 1 d after the last treatment.

FITC-conjugated Abs to CD3 (145-2C11), CD8 (53-6.7), CD11c (M1/70), CD19 (6D5), Gr-1 (RB6-8C5), TCRβ (h57-597), TCRγδ (UC7-13D5); Alexa 488–conjugated Abs to CD103 (2E7); PE-conjugated Abs to CD3 (145-2C11), CD19 (6D5), I-A (MHC class II) (M5/114.15.2); PE-Cy7–conjugated Abs to CD11b (M1/70); allophycocyanin-conjugated Abs to CD11c (N418), CD25 (PC61); CD117 (2B8), KLRG1 (2F1); Alexa 647–conjugated Abs to CD45.2 (104); APC-Cy7–conjugated Abs to CD4 (GK1.5); and streptavidin and Brilliant Violet 510–conjugated Abs to Thy1 (53-2.1) and Sca1 (D7) were purchased from BioLegend. FITC-conjugated Abs to B220 (RA3-6B2) and NK1.1 (PK136); PE-conjugated Abs to CD135 (A2F10); PE-Cy7–conjugated Abs to CD45.1 (A20) and CD127 (A7R34); eFluor 660–conjugated Ab to NKp46 (29A1.4), and biotinylated Ab to Sca1 (D7) were purchased from eBioscience. Brilliant Violet 421–conjugated Ab to CD117 (2B6) was purchased from BioLegend or BD Biosciences. PE-conjugated Ab to α4β7 (DATK32) was purchased from BD Biosciences. The lineage mixture for ILC identification consisted of Abs to CD3, CD8, CD11c, CD19, B220, NK1.1, Gr-1, TCRβ, and TCRγδ. For intracellular staining, the Foxp3 staining buffer kit with PE- or PerCP-eFluor 710–conjugated Ab to RORγt (AFKJS-9) and eFluor 660–conjugated Ab to Gata3 (TWAJ, eBioscience) was used. For IL-22 detection, cells were stimulated with 50 ng/ml PMA and 500 ng/ml ionomycin (PMA/Iono) or 20 ng/ml IL-23 in IMDM containing 5% FCS for 4 h. For the last 2 h, brefeldin A (10 μg/ml) was added. Cells were fixed with 4% paraformaldehyde, permeabilized in 0.5% saponin, and stained with PE-conjugated IL-22 (1H8PWSR; eBioscience). For RORγt staining, cells were refixed with Foxp3 staining buffer kit. Dead cells were excluded with the Fixable Viability Dye eFluor 450 (eBioscience). Cells were acquired using an FACS Canto II or LSR Fortessa (BD Biosciences) and analyzed with Flow Jo software (Tree Star). Cell sorting was done using an FACS Aria II (BD Biosciences).

For detection of PP anlagen, whole-mount VCAM staining of SIs from 1- to 2-d-old mice was performed as previously described (18). Briefly, after removing the serosa, intestines were fixed with 4% paraformaldehyde. Free aldehyde groups were quenched with 4% glycin in PBS and tissue was rehydrated using 50, 70, and 100% methanol. Endogenous peroxidase was blocked by incubation with 30% H₂O₂ in methanol. After blocking with PBS containing 1.5% skim milk and 0.1% Triton X-100, intestines were incubated with biotinylated anti-VCAM Ab (clone 429; eBioscience) overnight at 4°C. After extensive washing with PBS containing 1.5% skim milk and 0.1% Triton X-100, intestines were incubated with HRP-conjugated streptavidin (BioLegend) at room temperature. After washing with PBS and TBS, 3,3′-diaminobenzidine substrate (Sigma-Aldrich) was added to visualize Ab binding.

BM cells were obtained by crushing bones in a mortar. Mesenteric tissue of neonatal mice, FL of embryonic day (E) 14.5 embryos and adult spleen were smashed between two glass slides. For ILC detection, all organs were digested in DMEM with 0.025 mg/ml DNAseI (Roche) and 1 mg/ml Collagenase D (Roche) at 37°C on a rocking platform, and cells were collected in DMEM, 5% FCS, and 2 mM EDTA. For detection of splenic and mesenteric ILCs, organs from adult mice were cut into small pieces and digested three times for 15 min. The lung was cut into pieces, transferred into gentleMACS C-tubes containing 5 ml digestion medium, and stirred on the gentleMACS Octo Dissociator (Miltenyi Biotec), incubated for 30 min, and subsequently stirred again. Cells were passed through a 70-μm filter. Lamina propria preparation from SI of adult mice was done as described previously (21). Briefly, SIs were collected, PPs were removed, and SIs were opened longitudinally, cut into 1-2 cm pieces, and incubated in Ca²⁺- and Mg²⁺-free PBS containing 30 mM EDTA for 30 min on ice. Epithelial and intraepithelial cells were removed by shaking vigorously and repeated washing of the tissue with PBS. Intestinal pieces were digested four times for 15 min. Lamina propria cells were purified using a Percoll (GE Healthcare) gradient at 20°C and 1800 rpm for 30 min. Cells of the interphase were collected. SI from 0.5-d-old mice were collected, and mesenteric tissue was removed under the microscope. Intestines were opened longitudinally, washed with PBS by vigorous shaking, cut into small pieces, and digested two times for 30 min. E15.5 intestines were collected under the microscope, and three intestines were pooled. Intestines were cut into small pieces and digested two times for 30 min.

For competitive BM transfer, CD45.1⁺/CD45.2⁺ WT mice were lethally irradiated and reconstituted with 1 × 10⁷ BM cells in a 1:1 or 1:5 ratio of CD45.1⁺ WT BM and CD45.2⁺flt3l^−/− BM cells. For competitive transfer of CLPs, lineage⁻ (Lin⁻)CD117^lowCD127⁺Sca1^lowCD135⁺ CLPs from CD45.1⁺ WT mice, CD45.2⁺ WT mice, and CD45.2⁺flt3l^−/− mice were sorted. A total of 1000 CD45.1⁺ WT CLPs was mixed either with 1000 CD45.2⁺flt3l^−/− CLPs or with 1000 CD45.2⁺ WT CLPs and transferred into Rag2^−/−Il2rg^−/− mice (CD45.2⁺).

For differentiation of CLPs into ILCs, CLPs were sorted from the BM of Rag2^−/− mice. A total of 100–300 cells/well was plated on OP9 stromal cells in 96-well flat-bottom plates and cultured for 10–14 d in IMDM and 10% FCS supplemented with 5% IL-7 supernatant (SN), 5% IL-2 SN, and 20 ng/ml SCF. Additionally, Flt3L was added in concentrations of 25, 50, or 100 ng/ml. For survival and proliferation of intestinal ILCs, ILC2s and ILC3s were sorted from the SI of Rag2^−/− mice and labeled with 7.5 μM CFSE. A total of 30,000 cells/well was seeded in 96-well U-bottom plates and cultured for 4 d in IMDM, 10% FCS without cytokines, with 20 ng/ml IL-7, 100 ng/ml Flt3L, or 20 ng/ml IL-7 plus 5% IL-2 SN. For detection of apoptotic and necrotic cells, ILCs were stained with Annexin V and 7-aminoactinomycin D (7-AAD).

Statistical analysis was performed with GraphPad Prism 6 for Macintosh OSX (GraphPad Software). All data were tested for normal distribution using the Shapiro–Wilk test. If data were normally distributed, differences between groups were calculated using a two-tailed unpaired Student t test. If data were not normally distributed, differences between groups were calculated using the Mann–Whitney U test. Statistically significant differences are depicted as follows: *p < 0.05, **p < 0.01, ***p < 0.001.

During fetal development, LTi cells can arise from CD117⁺CD127⁺α4β7⁻ (33) or CD117⁺CD127⁺α4β7⁺ (32) FL progenitors, which were reported to express CD135 (34, 36). Accordingly, we found that the majority of both α4β7⁻ and α4β7⁺Thy1⁻ progenitors in the FL were positive for CD135 (Supplemental Fig. 1A). We have previously shown that IL-7 is crucial for LTi cell (45) and LN development (31) and that, in addition to IL-7, other cytokines promote intestinal LTi cell and PP development (29, 31). To investigate whether Flt3L has an influence on LTi cell development, we analyzed fetal and neonatal flt3l^−/− mice and compared them with WT and Il7^−/− mice. In the FL, α4β7⁻ and α4β7⁺ progenitors were significantly reduced in E14.5 flt3l^−/− mice compared with WT and Il7^−/− mice (Fig. 1A). In neonatal flt3l^−/− mice, PP development was greatly impaired (Fig. 1B). Accordingly, we detected a 3-fold reduction of LTi cells in the SI of neonatal flt3l^−/− mice (Fig. 1C). It was reported that a CD11c^low lymphoid cell population called lymphoid tissue-initiating cell (LTin) contributes to PP development (46). Although LTin cells did not express flt3 (Supplemental Fig. 1B) and were found in normal frequencies in the intestine of E15.5 flt3l^−/− mice, the absolute LTin cell number was significantly reduced (Fig. 1D), suggesting that they originate from an Flt3L-dependent progenitor cell.

In contrast to PPs, we could detect almost all LNs (Fig. 1E) together with normal LTi cell numbers in the mesenteries of neonatal flt3l^−/− mice (Fig. 1F). In comparison, Il7^−/− mice had reduced LTi cell numbers (Fig. 1F), and LN development was impaired (Fig. 1E). Because LTi cells were negative for flt3 (Supplemental Fig. 1B, 1C), we conclude that Flt3L deficiency mainly affected LTi progenitors in the FL.

To investigate the role of Flt3L for adult ILCs, we compared the number (Fig. 2B) and frequency (Fig. 2C) of NK cells, ILC2s, and ILC3s in the SI of WT, Il7^−/−, flt3l^−/− mice, and in flt3l-tg mice overexpressing Flt3L under the control of the human β-actin promoter (42). The gating strategy of the different ILCs is shown in Fig. 2A. As already reported, NK cell numbers were normal in the SI of Il7^−/− mice, whereas ILC2s and ILC3s were reduced (Fig. 2B, 2C) (13, 15). In contrast, loss of Flt3L led to a considerable reduction of all types of ILCs in the SI compared with WT controls. Deletion of Flt3L had no influence on the function of ILC3s, because stimulation with PMA/Iono or IL-23 led to normal production of IL-22 in flt3l^−/− mice (Supplemental Fig. 2). The transgenic overexpression of Flt3L had the opposite effect, leading to an enormous increase in NK cell, ILC2, and ILC3 numbers (Fig. 2B) and a slightly elevated frequency of IL-22⁺ ILC3s after PMA/Iono stimulation (Supplemental Fig. 2).

ILC3s can be subdivided into several subsets according to their expression of CD4 and NKp46 (21, 22) (Fig. 2A). Because Flt3L could differentially affect these ILC3 subsets in the SI, we analyzed CD4⁺ ILC3s, NKp46⁺ ILC3s, and CD4⁻NKp46⁻ (double-negative [DN]) ILC3s. Although the frequency of the different ILC3 subsets was not changed in flt3l^−/− mice, we detected a 4-fold reduction of DN and NKp46⁺ ILC3s and a 8-fold reduction of CD4⁺ ILC3s compared with WT mice (Fig. 2D). Conversely, in flt3l-tg mice, CD4⁺ ILC3s were increased 20-fold, DN ILC3s 16-fold, and NKp46⁺ ILC3s 9-fold compared with WT controls (Fig. 2D). This led to a higher frequency of CD4⁺ ILC3s and a reduced frequency of NKp46⁺ ILC3s in flt3l-tg mice (Fig. 2D).

For assessing whether Flt3L also affected ILC numbers in other organs, we analyzed the number (Fig. 2E, 2F) and frequencies (Fig. 2G) of ILC2s and ILC3s in the mLN and spleen as well as ILC2s in the lung. The overexpression of Flt3L significantly increased ILC2 numbers in mLN, spleen, and lung. Deletion of Flt3L resulted in reduced numbers of ILC2s only in the lung and mLN. In contrast, IL-7 deficiency led to a severe reduction of ILC2s in all three organs (Fig. 2E). As observed for ILC2s, transgenic overexpression of Flt3L led to an increase of ILC3s in mLN and spleen (Fig. 2F). Loss of Flt3L led to a significant reduction of ILC3s in mLN and spleen, whereas IL-7 only affected ILC3 numbers in the mLN. Taken together, Flt3L influences ILC numbers in various organs of adult mice, though with some differences in the magnitude.

The results obtained from flt3l-tg mice prompted us to ask whether the treatment of adult mice with rFlt3L could increase ILC numbers in WT mice and restore ILC numbers in Il7^−/− and flt3l^−/− mice. Accordingly, we injected rFlt3L over a period of 10 d into adult mice. As a control, we determined the number of CD11c^high dendritic cells (DCs) that are known to expand during Flt3L administration (47). DC numbers in the SI were highly increased after Flt3L treatment in all three strains (Fig. 3A). In WT mice, the number of NK cells and ILC2s was significantly higher in Flt3L-treated mice compared with PBS-treated controls (Fig. 3B). Among ILC3s, the CD4⁺ and DN subsets mainly responded to Flt3L treatment in WT mice (Fig. 3C), reminiscent of our previous data from flt3l-tg mice. In Il7^−/− mice, NK cells and ILC3s expanded upon Flt3L treatment, whereas ILC2s were not changed in numbers (Fig. 3B, 3C). In contrast, Flt3L treatment was unable to expand the number of NK cells, ILC2s, and ILC3s in flt3l^−/− mice (Fig. 3B, 3C). Together, these data confirm our previous findings that Flt3L promotes ILC3 development preferentially affecting CD4⁺ and DN ILC3 subsets. In addition, we demonstrate that Flt3L treatment is sufficient to expand ILC3s even in the absence of IL-7.

Flt3L is an important cytokine for the development of DCs (41, 48). In the intestine, three subsets of DCs can be discriminated by their expression of CD103 and CD11b (49). CD103⁺ DCs were described to be responsive to Flt3L (49). Indeed, CD11c^high DCs, which contain mainly CD103⁺ DCs, were dramatically reduced in the SI of flt3l^−/− mice, whereas numbers of CD11c^low cells, including CD103⁻ DCs and macrophages (MΦ), were comparable to WT mice (Fig. 4A, 4B). Recently, it has been reported that CX₃CR1⁺ DCs can regulate ILC3 numbers in the SI (50). To investigate whether the effect of Flt3L on ILCs was indirectly mediated by DCs, we analyzed mice in which DCs were constitutively ablated by crossing CD11c-Cre mice with mice expressing diphtheria toxin A (DTA) under the control of a loxP flanked stop cassette in the Rosa26 locus (R-DTA). The resulting CD11c-Cre/R-DTA (ΔDC) mice lack >90% of conventional DCs including myeloid, lymphoid, and plasmacytoid DCs in thymus, spleen, and LNs (51). In the SI, ΔDC mice showed a severe reduction of CD11c^high DCs with an almost complete absence of CD103⁺ DCs (Fig. 4A, 4C). In contrast, CD11c^low MΦ were only 2-fold reduced (Fig. 4A, 4C). Despite the fact that ΔDC mice and flt3l^−/− mice both had reduced DC numbers in the SI, ΔDC mice had normal numbers of PPs (Fig. 4D), suggesting that DCs are not required for LTi cell development. Whereas NK cell numbers were normal in the SI of ΔDC mice, ILC2 numbers were increased (Fig. 4E). Analysis of the different ILC3 subsets revealed that the number of NKp46⁺ ILC3s but not CD4⁺ or DN ILC3s was elevated (Fig. 4F). In order to test whether exogenous administration of Flt3L could increase ILC numbers in the absence of DCs, we treated ΔDC mice and their littermates for 10 d with rFlt3L. Flt3L administration led to a striking increase of DCs in the SI of litters and expansion of only few CD11c^high DCs in ΔDC mice (Fig. 4G). In line with our previous data, we observed a significant increase of CD4⁺ ILC3s and higher numbers of DN ILC3s in Flt3L-treated ΔDC mice (Fig. 4H). Altogether, our data indicate that the effect of Flt3L on ILCs is not mediated by DCs.

In the adult SI, we neither found Flt3 protein nor transcript expression in ILC2s or ILC3s (Supplemental Fig. 1D, 1E), indicating that Flt3L does not act directly on mature ILCs. To confirm this, we tested the role of Flt3L on the survival and proliferation of intestinal ILC2s and ILC3s. Therefore, we sorted ILC2s and ILC3s from the SI of Rag2^−/− mice, labeled them with CFSE, and cultured them for 4 d without cytokines, with Flt3L or IL-7 alone or with a combination of IL-7 and IL-2. Using Annexin V and 7-AAD staining, we observed that both ILC2s and ILC3s hardly survived without addition of cytokines or with Flt3L alone (Fig. 5A). In contrast, addition of IL-7 alone or in combination with IL-2 supported survival of both cell types (Fig. 5A). Although ILC2s and ILC3s proliferated with IL-7 alone or IL-7 plus IL-2, Flt3L had no influence on proliferation of mature ILC2s or ILC3s (Fig. 5B). Our data indicate that Flt3L is dispensable for survival or proliferation of mature intestinal ILCs.

Because ILCs did not respond directly to Flt3L, we focused on the effect of Flt3L on ILC progenitors in the BM. Therefore, we analyzed the number of Lin⁻Sca1⁺c-Kit^high cells (LSKs) and CLPs as early lymphoid progenitors (Fig. 6A). In addition, we determined the number of progenitors that were committed to the ILC lineages; firstly, the recently described common progenitor to all helper-like ILCs (CHILP), which gives rise to all ILCs except NK cells (9), and secondly, the ILC2 progenitor (ILC2P) (15) (Fig. 6A). Analysis of CD135 expression on these BM progenitors showed that the majority of LSKs and all CLPs were positive for CD135, whereas CHILPs and ILC2Ps were negative (Supplemental Fig. 1F). In line with a previous report (52), we found normal numbers of LSKs in flt3l^−/− mice, whereas CLP numbers were reduced (Fig. 6A). In Il7^−/− mice, LSKs and CLPs were present in normal numbers (Fig. 6A). Similarly, CHILP numbers were normal in Il7^−/− mice but considerably reduced in flt3l^−/− mice compared with WT controls (Fig. 6A). In sharp contrast to CLPs and CHILP, deletion of Il7 led to a severe loss of ILC2Ps, whereas these cells were less affected by the lack of Flt3L (Fig. 6A).

Although CLP numbers were reduced in flt3l^−/− mice, not all CLP-derived lymphocytes were affected to the same degree in the periphery. In the spleen, NK cells were severely reduced, whereas B cells were only diminished 2-fold, and T cell numbers did not differ between WT and flt3l^−/− mice (Fig. 6B). In the SI, no significant difference in IgM⁺ B cell and T cell numbers was detectable in WT and flt3l^−/− mice (Fig. 6B). To investigate whether Flt3L promotes ILC3 development, we sorted CLPs from the BM of Rag2^−/− mice and cultured them for 10–14 d on OP9 stromal cells in the presence of IL-7, SCF, and IL-2. Additionally, Flt3L was added in different concentrations. Without Flt3L, CLPs predominantly gave rise to CD19⁺ pro-B cells. In contrast, addition of Flt3L in all concentrations supported RORγt⁺ ILC3 development (Fig. 6C). The striking effect of Flt3L deficiency on ILCs but not on B and T cells and the data obtained from CLP differentiation prompted us to ask whether CLPs from flt3l^−/− mice have an intrinsic defect in generating ILCs. To test this, we performed competitive BM transfers by reconstituting lethally irradiated CD45.1⁺/CD45.2⁺ WT mice with BM from CD45.1⁺ WT and CD45.2⁺ flt3l^−/− mice (Fig. 6D). Six weeks after reconstitution, we analyzed the frequency of donor-derived cells within SI ILC3s. When we mixed the BM cells in a 1:1 ratio, we observed that 75% of ILC3s were derived from WT and 25% from flt3l^−/− BM (Fig. 6D). As CLP numbers are reduced in flt3l^−/− mice (Fig. 6A), we decided to compensate this disadvantage of low progenitor frequency by using four times more BM cells of flt3l^−/− mice than of WT mice. The BM ratio of 1:5 (WT/knockout) was assumed to adjust the CLP ratio to 1:1. With this approach, the frequency of WT- and flt3l^−/−-derived SI ILC3s was 50% each (Fig. 6D), indicating that CLPs from flt3l^−/− mice have the same capacity to reconstitute ILCs as WT CLPs. To confirm this, we adoptively transferred FACS-sorted CLPs from CD45.1⁺ WT mice and from CD45.2⁺flt3l^−/− mice in a 1:1 ratio into Rag2^−/−Il2rg^−/− mice (CD45.2⁺). As controls, we reconstituted Rag2^−/−Il2rg^−/− mice with a 1:1 mixture of CD45.1⁺ WT CLPs and CD45.2⁺ WT CLPs (Fig. 6E). After 6 wk, the frequency of ILC3s derived from WT and flt3l^−/− CLPs was comparable (Fig. 6E), showing that flt3l^−/− CLPs had no intrinsic defect in ILC lineage commitment.

Flt3L plays an essential role in survival of lymphoid progenitors in the BM (52) and differentiation of DCs (41, 48). Our study extends these data to a role for Flt3L in the development of LTi cells and PPs during fetal life by affecting the ILC progenitor pool in the FL. We also identify Flt3L as important factor regulating ILC progenitor numbers and consequently sufficient ILC numbers in adult mice. Finally, our data suggest that Flt3L and IL-7 act on different stages of ILC development.

The development of LNs and PPs depends on LTi cells (7) and is regulated by cytokines such as IL-7, SCF, and receptor activator of NF-κB ligand (29, 53–55). Our data demonstrate that Flt3L deficiency caused a significant reduction in FL LTi progenitor cells, intestinal LTi cells, LTin cells, and PPs. Despite reduced LTi progenitor cell numbers, LN development and LTi cell numbers in mLNs were relatively normal. Similar results were observed in Nfil3^−/− mice (56), in which LNs but not PPs develop despite reduced ILC progenitor numbers (57–59). One explanation for this might be that IL-7, an important growth and survival factor for LTi cells, is poorly expressed in the fetal intestine but is abundant in fetal LN anlagen (29). Therefore, LN-derived IL-7 may potentially compensate the low LTi cell progenitor numbers in flt3l^−/− mice. In contrast, the lack of PPs is presumably caused by the simultaneous reduction of both LTin and LTi cell numbers in the SI of flt3l^−/− mice.

In the adult system, we could show that Flt3L deficiency resulted in a severe reduction of ILC progenitor cells in the BM and mature ILCs in various organs including intestine, spleen, lung, and mLN. The strongest effect was seen in the intestine and the intestine-associated mLN. We cannot fully exclude that some differences in the degree of ILC reduction are a consequence of additional local factors facilitating ILC recruitment, survival, or expansion. Currently, it is still not clear how the different ILC3 subsets are developmentally related. In our study, we could observe that in the intestine CD4⁺ ILC3s and DN ILC3s were mainly affected by the lack or the increased availability of Flt3L. Whether progenitors for ILC3s can exploit other cytokines to generate NKp46⁺ ILC3s or whether each ILC3 subset originates from distinct intermediate progenitors is currently unknown. The latter possibility is corroborated by the finding that CD4⁺ ILC3s can only be generated from PLZF⁻ CHILPs (60).

In addition to the Flt3L transgenic mouse model, we also observed an increase of ILCs in WT mice after administration of rFlt3L, which was previously shown to increase the number of DCs (47). Also, NK cells and regulatory T cells were reported to expand after Flt3L treatment (61–65). The increase of these two cell types was shown to be a result of the Flt3L-driven expansion of DCs and their secretion of IL-2 and IL-15 (62–65). As the expansion of DCs in flt3l-tg mice or after Flt3L treatment could be responsible for the observed increase of ILC2s and ILC3s, we analyzed ΔDC mice, which lack CD11c^high DCs. The normal or even increased ILC numbers seen in ΔDC mice under steady-state conditions and the increase of CD4⁺ and DN ILC3s after Flt3L administration in these mice strongly argues against the hypothesis that ILC expansion is driven by DCs. The inability of Flt3L treatment to increase ILCs in flt3l^−/− mice, albeit increasing DC numbers, also supports this conclusion. It was recently shown that the number of NKp46⁺ ILC3s depends on CX₃CR1⁺ mononuclear cells (50), which contain CD103⁻ DCs and MΦ (66). In ΔDC mice, NKp46⁺ ILC3s were slightly increased, which might be explained by the relative enrichment of CD11c^lowCD103⁻ MΦ in these mice.

In this study, we could show that Flt3L does not act directly on mature ILCs, because neither ILC2s nor ILC3s showed enhanced survival or proliferation with Flt3L. Instead, the effect of Flt3L on ILCs was mediated by the regulation of ILC progenitor numbers in the BM. The more restricted ILC progenitor, the CHILP, was also dramatically reduced in flt3l^−/− mice, even though this progenitor does not express CD135 and therefore must be considered as unresponsive to Flt3L. This suggests that the reduction of CLP numbers in flt3l^−/− mice directly affects the number of downstream progenitors and that the loss of CLPs cannot be compensated by other cytokines during the transition to CHILP. Although reduced CLP numbers in flt3l^−/− mice predominantly resulted in a loss of ILCs, competitive transfer experiments showed that CLPs from flt3l^−/− mice have the same ability as WT CLPs to give rise to ILC3s. In addition to CLP development, the in vitro data point to a beneficial role for Flt3L in the generation of ILC3s. Whether this is also the case for ILC2s or NK cells is not clear, as our culture condition did not support the development of these cells.

In contrast to the striking reduction of CLP numbers, ILC2P numbers were only moderately reduced in flt3l^−/− mice, suggesting that ILC2Ps are able to expand in flt3l^−/− mice possibly as a consequence of responsiveness to IL-7. Together with the data from Klose et al. (9), it can be assumed that the ILC2P is the first ILC-committed progenitor that is affected by the loss of IL-7R signaling. The importance of IL-7 for the development of ILC2s is also reflected by the fact that Il7^−/− mice had fewer ILC2s than flt3l^−/− mice and that Flt3L treatment was not able to increase ILC2 numbers in Il7^−/− mice. Whether IL-7 is also important during the transition of CHILP to ILC3s or whether IL-7 only promotes ILC3 survival in the periphery remains to be investigated.

The unequal number of ILC progenitors in WT, Il7^−/−, and flt3l^−/− mice might explain why Flt3L treatment was not able to increase ILC numbers in flt3l^−/− mice. Thus, it is likely that an extended treatment with Flt3L can compensate for low progenitor numbers and can consequently increase ILC numbers in flt3l^−/− mice. In addition, we cannot exclude that flt3l^−/− mice mounted an immune response against the recombinant protein, which led to the neutralization and subsequently lower abundance of Flt3L.

Taken together, our study demonstrates that the presence of Flt3L is important to generate ILCs during fetal development as well as during adult life. We could show that IL-7 and Flt3L are both important for ILC numbers in adult mice; however, our data suggest that they operate at different developmental stages. Flt3L mainly acts on CLPs and CHILPs, whereas the effect of IL-7 occurs at later stages of ILC development, namely on ILC2Ps and mature ILC2s and ILC3s. In addition, our data offer the possibility to use Flt3L as therapeutic approach for restoring intestinal ILCs in patients with SCID and for immune protection of mucosal surfaces. Because Flt3L treatment is already approved for treatment of cancer patients (67), beneficial effects of this cytokine for mucosal immunity could be tested in human trials.

We thank T. Barthlott for cell sorting; S. Sawa for protocols; and S. Eckervogt, E. Terszowska, L. Jäckel, and R. Recinos for animal work. We also thank R. Ceredig, R. Tussiwand, and M. Gaio for critical reading of the manuscript.

The authors have no financial conflicts of interest.