Adaptation of Innate Lymphoid Cells to a Micronutrient Deficiency Promotes Type 2 Barrier Immunity

See allHide authors and affiliations

Science  24 Jan 2014:
Vol. 343, Issue 6169, pp. 432-437
DOI: 10.1126/science.1247606


How the immune system adapts to malnutrition to sustain immunity at barrier surfaces, such as the intestine, remains unclear. Vitamin A deficiency is one of the most common micronutrient deficiencies and is associated with profound defects in adaptive immunity. Here, we found that type 3 innate lymphoid cells (ILC3s) are severely diminished in vitamin A–deficient settings, which results in compromised immunity to acute bacterial infection. However, vitamin A deprivation paradoxically resulted in dramatic expansion of interleukin-13 (IL-13)–producing ILC2s and resistance to nematode infection in mice, which revealed that ILCs are primary sensors of dietary stress. Further, these data indicate that, during malnutrition, a switch to innate type 2 immunity may represent a powerful adaptation of the immune system to promote host survival in the face of ongoing barrier challenges.

An Immune Response to Malnutrition

Mucosal surfaces, such as those lining the intestine, are in constant contact with potentially pathogenic microbes, including bacteria and parasitic worms. This necessitates so-called barrier immunity, which is mediated in part by innate lymphoid cells, subsets of which combat specific types of infection. Although malnutrition has been associated with immunosuppression, Spencer et al. (p. 432) now show that vitamin A deficiency selectively activates one branch of barrier immunity. Vitamin A deficiency in mice enhanced immunity to chronic worm infections by increasing the levels of one subset of innate lymphoid cells lacking the corresponding retinoic acid receptor. In contrast, another innate lymphoid cell subset that carries the vitamin A receptor and is important for bacterial immunity was depleted. Thus, the immune system can adapt its response to dietary stress, thereby promoting host survival.

Maintenance of barrier defense is an absolute requirement for mammalian host health and survival. Intestinal immunity has evolved in the context of constitutive exposure to microbial pressures. In this regard, the gastrointestinal (GI) tract is home to an estimated 100 trillion commensals, and more than 2 billion humans are chronically infected with parasitic worms (1, 2). Over the course of evolution, maintenance of barrier immunity had to adapt to unstable nutritional availability in the face of these ongoing challenges. Although dietary ligands can be sensed by immune cells (36), it is not clear how the immune system integrates dietary cues in order to tune immune responses on the basis of the nutritional state of the host.

Malnutrition remains the primary cause of immunosuppression worldwide (7). Nonetheless, despite profoundly impaired adaptive immunity associated with malnutrition, most humans can survive for extended periods of severe dietary restriction. We postulated that compensatory mechanisms might be in place to sustain defined branches of immunity and, in particular, responses associated with the protection of barrier tissues. Vitamin A deficiency is one of the most common nutrient deficiencies, affecting an estimated 250 million children in regions of the world where chronic worm infections are prevalent (8). This essential micronutrient supports adaptive immunity through its metabolite, retinoic acid (RA), which is highly enriched in the gastrointestinal tract (9, 10). Here, we examine the possibility that innate lymphoid cells (ILCs)—potent mediators of barrier maintenance, tissue repair, and host defense (11, 12)—may be primary sensors of dietary stress able to sustain barrier defense in the context of vitamin A deficiency.

Several subsets of ILCs have been described, with two dominant tissue-resident populations (12). At steady state, retinoic acid receptor (RAR)–related orphan receptor gamma (RORγt+ ILCs) (ILC3s) are dominantly present in the gut, whereas the majority of ILCs residing in the lung or skin belong to the GATA3+ ILC2 subset (11, 13, 14) (Fig. 1, A and B). We explored the possibility that vitamin A metabolites may act as a local cue to control ILC populations. As previously described, T helper cells, TH1 and TH17 (15, 16), as well as TH2 cells, were significantly reduced in the gut of vitamin A–deficient (VAI) mice (fig. S1, A and B). The number of gut resident ILCs, ILC1s, and lymphoid tissue inducer cells (LTi’s) remained unchanged in VAI mice (fig. S2, A to D). However, ILC3s and ILC-derived IL-22 and IL-17 were greatly reduced in VAI wild-type (WT) and Rag1–/– mice (Fig. 1, C to H, and figs. S3 and S4). Impaired cytokine expression was associated with reduced expression of RORγt (fig. S4B). In contrast, we observed a significant increase in ILC2- and ILC-derived IL-13, IL-5, and IL-4 in the gut of VAI mice (Fig. 1, F, G, and H, and fig. S5, A and B).

Fig. 1 ILC3s are enriched in the GI tract and depend on RA.

(A) Flow cytometric analysis of cells isolated from small intestinal lamina propria (gut), lung, and skin of naïve C57BL/6 mice. (Top) Live CD45+ cells stained with Thy1.2 and lineage (lin) markers (NK1.1, T cell receptors TCRβ and TCRγδ, CD11b, CD11c, CD4, CD8α, CD8β, CD19, GR-1, DX5, and Ter119). (Bottom) Cells gated on Lin and Thy1.2 expression (ILCs), stained for RORγt (ILC3s) and GATA3 (ILC2s). (B) Frequencies of ILC2s and ILC3s in gut, lung, and skin. (C) Small intestinal lamina propria (SiLP) cells from control (Ctrl) or vitamin A–insufficient (VAI) WT or Rag1–/– mice, gated on Lin Thy1.2+ cells and analyzed for GATA3 and RORγt expression. (D and E) Total numbers of ILC3s (RORγt+) and ILC2 (GATA3+) cells in the SiLP of WT and Rag1–/– mice. (F) Intracellular IL-13 and IL-22 expression in Lin Thy1.2+ cells after stimulation with phorbol 12-myristate 13-acetate (PMA) and ionomycin. (G and H) Total numbers of IL-22– and IL-13–producing ILCs in the SiLP. (I) SiLP ILC2s and ILC3s from Rag1–/– mice treated with vehicle control (Veh) or RAi for 8 days. (J) Intracellular IL-13 and IL-22 expression in ILCs after stimulation with PMA and ionomycin. Results are representative of at least three independent experiments with three to five mice in each experimental group. All graphs display means ± SEM.

Acute inhibition of RA signaling with the pan-RAR inhibitor BMS493 (RAi) (17) also resulted in reduced ILC3s and ILC-derived IL-22 and inversely increased ILC2s and ILC-derived IL-13 in both WT and Rag1–/– mice (Fig. 1, I and J, and fig. S6, A to D). Alterations in ILC subsets subsequent to RA deprivation occurred independently of commensals (fig. S7). Short-term treatment of VAI Rag1–/– mice with RA restored ILC3 frequencies and cytokine production to levels found in control mice and reduced ILC2 numbers (Fig. 2, A to C). VAI Rag1–/– mice or mice treated with RAi displayed lower frequencies of Ki67-expressing ILC3s than control mice, whereas the number of Ki67-expressing ILC2s was greatly increased (Fig. 2, D and E). Conversely, addition of RA reversed the proliferative potential of ILCs to frequencies observed in control mice (Fig. 2D). Thus, RA acts as a switch to control a proliferative balance between the two intestinal ILC subsets.

Fig. 2 RA dynamically regulates developmental balance between ILC subsets.

SiLP cells from Ctrl and VAI or VAI Rag1–/– mice treated with all-trans RA every 3 days for 12 days (VAI + RA). (A) GATA3 and RORγt expression in Lin Thy1.2+ cells (top). Intracellular IL-13 and IL-22 expression in Lin Thy1.2+ cells after stimulation with PMA and ionomycin (bottom). (B) Total numbers of ILC3s and ILC2s in SiLP and (C) total numbers of IL-22– and IL-13–producing ILCs in the SiLP. (D) Frequencies of Ki67 expression in ILC3s and ILC2s isolated from SiLp of Ctrl, VAI, or VAI +RA Rag1–/– mice. (E) SiLP cells from Rag1–/– mice treated with RAi or Veh for 8 days stained for intracellular Ki67 in ILC3s (top) cells and ILC2 (bottom). (F) CD45.1+ CLPs (n = 100,000) were transferred into congenic CD45.2+ Rag2–/–γc–/– mice and treated either with Veh, RA, or RAi. Fourteen days after transfer, SiLP cells were isolated, stained for GATA3 and RORγt, and gated on CD45.1+, Lin and Thy1.2+ cells. (G) Quantification of relative proportion of ILC3s and ILC2s in recipient mice and (H) frequencies of ILC2s and ILC3s from RARαfl/fl mice. Transferred cells were gated on green fluorescent protein–positive (GFP+) cells. (I) ILC2Ps were sort purified from bone marrow and cultured in vitro with IL-7 and SCF in the presence of Veh, RA, and RAi for 7 days. Total numbers of Thy1.2+ GATA3+ cells (left) and IL-13 production in the culture supernatant (right). (J) Ki67 expression in small intestinal ILC2s from mice treated with RAi or Veh and antibody against IL7α. (K) Total number of Ki67+ ILC2s and (L) total number of ILC2s. Data are representative of three (A to G) or two (H to L) independent experiments with three or four mice in each experimental group or at least two independent experiments with three experimental groups of cells isolated from two mice each (I). (G) displays pooled data from three independent experiments. All graphs display means ± SEM.

To address whether RA signaling influenced the fate of ILCs, we transferred common lymphoid progenitors (CLPs) to mice devoid of ILCs (Rag2–/–γc–/– mice) (12). Transferred CLPs gave rise to both ILC2s and ILC3s that accumulated in the GI tract (18, 19) (Fig. 2, F and G). Exogenous delivery of RA led to a dominant accumulation of ILC3s in the intestine, whereas inhibition of RA signaling favored ILC2 accumulation (Fig. 2, F and G). Both ILC2s and ILC3s selectively express the nuclear receptor RARα (20) (fig. S8). We retrovirally transfected progenitors with lymphoid potential obtained from mice carrying WT or floxed alleles of the RARα gene (RARαfl/fl) with the Cre recombinase (fig. S9). Transfer of control progenitors resulted in the preferential accumulation of ILC3s, whereas RARα-deleted progenitors predominately gave rise to ILC2s (Fig. 2H). Further, addition of RA impaired ILC2 development from ILC2 progenitors (ILC2Ps), whereas inhibition of RA signaling resulted in increased ILC2 accumulation and IL-13 production in culture (Fig. 2I). These results reveal a cell-intrinsic suppressive role for RA on ILC2 maturation.

We next postulated that some effects of RA may result from a direct action on mature ILCs. Highly purified ILC2s exposed to RA or ILC3s treated with RAi for several days maintained their phenotype; therefore, we argue against an interconversion of these two populations (fig. S10, A to D). As previously reported, RA directly increased IL-22 (20) and RORγt expression by mature ILC3s and promoted ILC3 accumulation (fig. S11, A to D). Conversely, RA had a negative impact on ILC2 fitness, whereas treatment with RAi increased their proliferation, numbers, and IL-13 production in both mouse (fig. S12, A to C) and human cells (fig. S13).

We next explored the possibility that vitamin A deficiency may tune the responsiveness of ILC2s to factors contributing to their survival or proliferation (11, 12). Using mice deficient in IL-25, thymic stromal lymphopoietin (TSLP), or IL-33 receptors allowed us to exclude a dominant role for these factors (figs. S14 and S15). Of note, expression of the IL-7Rα, required for ILC development and survival (11, 12), was reduced on ILC2s and ILC2Ps in the presence of RA, whereas the absence of RA signaling increased IL-7Rα expression by murine ILC2s and ILC2Ps, as well as human ILC2s (fig. S16, A to E). Increased IL7Rα expression conferred enhanced signaling capacity in response to IL-7 as measured by increased signal transducer and activator of transcription 5 (STAT5) phosphorylation (fig. S16F). Blockade of IL-7 signaling in vivo abrogated the increase in proliferation and accumulation of ILC2s after RA inhibition but not in control mice (Fig. 2, J to L, and fig. S17). Together these results demonstrate that vitamin A deficiency is associated with altered ILC homeostasis and, in particular, increased IL-13–producing ILC2s, an effect potentially mediated by increased IL-7 responsiveness.

A corollary of our findings is that differential levels of vitamin A could promote different classes of barrier immunity with physiological levels of vitamin A associated with functional ILC3 responses, whereas reduced levels were associated with enhanced innate type 2 immunity. Indeed, both WT VAI mice and mice treated with RAi displayed enhanced susceptibility and pathology to Citrobacter rodentium compared with control mice (Fig. 3, A and B), a phenotype associated with impaired TH17 and ILC3 responses (Fig. 3, C and D, and fig. S18, A to E). Treatment with RAi of Rag1–/– mice, in which ILCs are the dominant source of IL-22 (21), dramatically increased pathology and mortality after infection (Fig. 3, E to I), an effect reversed by exogenous delivery of IL-22 (Fig. 3, J and K). This observation may provide an additional explanation for the profound susceptibility to gastrointestinal bacterial infections observed in children suffering from vitamin A deficiency (22, 23).

Fig. 3 Vitamin A deficiency results in impaired immunity to bacterial infections.

WT (A to D) and Rag1–/– (E to K) mice treated with Veh or RAi were infected with C. rodentium. (A) Percentile change of original body weight and frequency of surviving animals and (B) colon length of WT mice treated with RAi or Veh. (C) Large intestine lamina propria (LiLP) cells isolated from Veh or RAi-treated WT mice 10 days after infection with C. rodentium, gated on CD4+ and TCRβ+ cells and analyzed for RORγt (top) and IL-22 expression (bottom). (D) Total numbers of RORγt+ and IL-22+ CD4+ T cells. (E) Percentile change of original body weight and frequency of surviving Rag1–/– mice treated with Veh or RAi and infected with C. rodentium. (F) Colon length and (G) representative hematoxylin and eosin (H&E)–stained histological sections of colonic tissue analyzed 10 days post infection; scale bars, 200 μm. (H) LILP ILC2s and ILC3s from Veh or RAi-treated Rag1–/– mice 10 days after infection with C. rodentium (top) and intracellular IL-17A and IL-22 expression in ILCs after stimulation with PMA and ionomycin (bottom). (I) Total numbers of RORγt-expressing ILCs and total numbers of IL-22–producing ILCs per colon. (J) Weight loss of Ctrl, VAI, or VAI Rag1–/– mice treated with IL-22 (VAI+IL-22) and infected with C. rodentium. (K) Colon length analyzed 14 days post infection. Data represent at least two independent experiments with three to five mice in each experimental group. All graphs display means ± SEM.

Our results, thus far, predict that although TH1, TH17, and ILC3 collapse under vitamin A deficiency, type 2 immunity might be paradoxically augmented via an enhanced ILC2 response. In support of this, withdrawal of vitamin A heightened mucus production (15). Goblet cells play a central role in barrier protection by the secretion of mucus and antimicrobial peptides (24). Consistent with increased ILC2s observed in the absence of vitamin A, VAI Rag1–/– mice displayed significant goblet cell hyperplasia and increased goblet cell–associated expression of the RELM-β gene, Retnlb, an effect largely dependent upon IL-13 (Fig. 4, A, B, and C). In these settings, IL-13 production was predominantly ILC2-derived (fig. S19). We next addressed whether enhanced ILC2 responses could compensate for the defect in TH2 immunity during vitamin A deficiency (2529). To this end, we used a high dose of Trichuris muris eggs associated with TH2 induction (30). Although blocking RA impaired TH2 induction, ILC2 numbers were significantly increased compared with those in infected control mice (fig. S20, A and B). Remarkably, RAi-treated mice controlled parasite burden comparably to control mice, which supports the idea that ILC2s sustained worm control (fig. S20C). Further, both VAI and RAi-treated mice showed enhanced protection to physiological low-dose T. muris, a response associated with significant increase in the numbers of ILC2s and IL-13–producing ILCs (Fig. 4, D to H and fig. S21, A and B). In agreement with the role of IL-13 in T. muris control (31), accelerated worm clearance was critically dependent on this cytokine (Fig. 4G). We next assessed the role of ILC2s in mediating worm expulsion independently of adaptive immunity. Remarkably, VAI Rag1–/– mice displayed enhanced protection and reduced worm burden associated with substantial increase in ILC2 numbers and IL-13 production (Fig. 4, H, I, and J). Thus, vitamin A insufficiency leads to sustained and, in some cases, augmented control of nematode infection via the promotion of ILC2-dependent type 2 immunity.

Fig. 4 Vitamin A deficiency increases ILC2-mediated immunity to helminth infections.

(A) Small-intestine histologic sections of Ctrl, VAI, or VAI Rag1–/– mice treated with antibody against IL-13 (VAI+αIL-13) stained with periodic acid–Schiff (PAS) to visualize goblet cells; scale bars, 200 μm. (B) Total numbers of PAS-positive goblet cells per crypt and (C) Retnlb (RELM-β) gene expression in the small intestine of Ctrl, VAI, or VAI+αIL-13 Rag1–/– mice. (D) Lamina propria ILC2s and ILC3s isolated from the cecum of Ctrl or VAI WT mice 13 days after oral infection with T. muris (top) and intracellular IL-13 and IL-22 expression in ILCs after stimulation with PMA and ionomycin (bottom). (E) Total numbers of ILC2s and (F) total numbers of IL-13–producing ILCs in the cecum. (G) Worm burden in the cecum of Veh, RAi, RAi-treated IL-13–/– (RAi IL-13–/–), and RAi mice treated with neutralizing antibody against IL-13 (RAi αIL-13) 12 days after infection. (H) Number of worms in Ctrl and VAi Rag1–/– mice 12 days after infection and (I) intracellular GATA3 and RORγt (top) and IL-13 and IL-22 expression (bottom) in cecal ILCs. (J) Total numbers of ILC2s and IL-13+ ILCs in the cecum of T. muris–infected mice. Data represent at least two (A and B and H to J) or three (D to G) independent experiments with three to five mice in each experimental group. Data in (G) Veh and RAi is pooled from three experiments, RAi αIL-13 and RAi IL-13–/– represent one experiment each. All graphs display means ± SEM.

Our work suggests that vitamin A and its metabolite RA function together as a dietary alarm signal that allows the host to immunologically respond to its nutritional state. Contrary to the current paradigm, we show that nutrient deficiency is not associated with global immunosuppression but rather can selectively activate a distinct arm of barrier immunity. Notably, we found that ILC2s act as primary sensors of dietary stress able to compensate for the collapse of adaptive immunity in settings of nutrient deprivation. Type 2 immunity and, in particular, IL-13 are associated with tissue repair, increased mucus production, and physiological responses all aimed at reinforcing barrier integrity and defense (32, 33). Furthermore, enhanced type 2 responses are clearly beneficial in the context of exposure to worms that have been partners throughout human evolution and still represent the major form of parasitic infection worldwide. Because nematodes compete with the host for nutritional resources, reinforcement of antihelminth immunity can also provide a substantial advantage to the host. Thus, in settings of malnutrition, a rapid switch to type 2 barrier immunity imposed by vitamin A deficiency may represent a powerful adaptation of the immune system to transiently promote host survival in the face of dominant barrier exposures. Such a strategy leaves the host vulnerable to potential encounters with acute diarrheal pathogens but could provide a survival strategy to temporarily reduce the pressure from its constitutive evolutionary partners, worms and commensals, in settings of nutritional deprivation.

Supplementary Materials

Materials and Methods

Figs. S1 to S22

Table S1

References (3439)

References and Notes

  1. Acknowledgments: This work was supported by the Division of Intramural Research of the National Institute of Allergy and Infectious Diseases (NIAID), NIH; Office of Dietary Supplements, NIH; NIH grant F30 DK094708 (S.P.S.); Human Frontier Science Program (C.W.); U.S. Department of Agriculture–Agricultural Research Service project plan #1254-32000-094-00D (J.U.); and by the Damon Runyon Cancer Research Foundation (Dale F. and Betty Ann Frey Fellow, J.A.H.). We thank the NIAID animal facility staff; K. Holmes and the NIAID sorting facility, in particular, C. Eigsti and E. Stregevsky; and K. Beacht and the NIAID gnotobiotic facility, in particular, C. Avecedo and D. Trageser-Cesler for technical assistance. We thank D. Artis for providing C. rodentium, W. J. Leonard for providing TSLPR–/– mice, and W. Paul for providing IL33R–/– mice. We thank the Belkaid lab for critical discussions regarding the manuscript. The data presented in this manuscript are tabulated in the main paper and the supplementary materials.
View Abstract

Navigate This Article