Cutting Edge: Innate Lymphoid Cells Suppress Homeostatic T Cell Expansion in Neonatal Mice

The immune system regulates homeostasis of lymphocytes by mechanisms that are not fully understood. In adult lymphopenic mice, naive T cells undergo homeostatic or lymphopenia-induced proliferation (LIP) in response to the recognition of self-peptide–MHC complexes (1). This process requires growth factors such as IL-7 (2), results in T cell activation, and often leads to tissue destruction (3). In contrast, LIP and its pathological consequences appear to be prevented in adult immunocompetent mice by several mechanisms, including regulatory T cells (4). Importantly, T cells continuously consume IL-7 and occupy self-peptide–MHC complexes (1). The competition for both homeostatic factors reduces their availability and prevents self-destructive T cell activation (2, 5). Therefore, the size of and the competition within the adult T cell pool are critical for the maintenance of immunological self-tolerance (5, 6).

Newborn T cell–competent mice are nearly devoid of peripheral T cells (7). Surprisingly, only a minority of thymic emigrants and adoptively transferred naive T cells undergo LIP in lymphopenic neonates (8–10), and lymphopenia-induced tissue destruction is not observed. Based on these findings, we hypothesized that T cell–independent mechanisms prevent LIP of neonatal T cells, subsequent autoaggression (3), and the generation of a limited T cell repertoire (11).

In this study, we show that IL-7R–dependent innate lymphoid cells (ILCs) suppress LIP and subsequent activation of CD8⁺ T cells in neonatal but not adult lymphopenic mice. This T cell–independent mechanism prevents spontaneous T cell activation and allows the development of a diverse naive T cell pool, which regulates T cell homeostasis and immunonological self-tolerance later in life.

C57BL/6J mice (B6; Thy1.2⁺), congenic B6.PL-Thy1^a/Cy mice (Thy1.1⁺), and C57BL/6-IL-7Rα^−/− (B6.129S7-Il7r^tm1lmx/J) and C57BL/6-JAK3^−/− (B6.129S4-Jak3^tm1Ljb/J) mice were purchased from Charles River Laboratories (Sulzfeld, Germany). Thy1.1⁺ OT-I mice (12) and C57BL/6-RAG2^−/− (RAG^−/−) mice were bred in our animal facility. All mice were kept under specific pathogen-free conditions at the central animal facility of the German Cancer Research Center. Animal care was approved by the regional regulating authorities.

The adoptive transfers of CD8⁺ T cells were performed as described previously (10). Briefly, CD8⁺ T cells from Thy1.1⁺ OT-I mice were purified using CD8α-specific microbeads and autoMACS (both Miltenyi Biotec, Bergisch Gladbach, Germany). For purification of naive, polyclonal CD8⁺ T cells expressing low levels of CD44 (CD44^low), splenocytes and lymph node cells from Thy1.1⁺ mice were incubated with a FITC-labeled mAb specific for CD44. Subsequently, anti-FITC microbeads and a CD8⁺ T cell isolation kit were used to deplete CD44^int/high and CD8⁻ cells by autoMACS (all Miltenyi Biotec). For labeling with CFSE (Molecular Probes, Eugene, OR), 5 × 10⁷/ml purified T cells were incubated with 7.5 μM CFSE in PBS for 20 min at 37°C, washed twice with ice-cold PBS, and resuspended in PBS prior to injection. CD8⁺ T cells (1 × 10⁶ to 4 × 10⁶) were injected i.p. into neonatal mice at their day of birth (day 1) or at days 2–3. Adult mice were injected i.v. At the indicated times, splenocytes from recipients were stained with mAbs for CD8α and Thy1.1 and analyzed individually using a FACSCalibur flow cytometer (Becton Dickinson). Specificity of staining was confirmed using isotype-matched control Abs.

Sixty to 80 μg/g body weight of the depleting Abs anti-Thy1.2 (30H12) or anti-CD4 (GK1.5) were injected i.p. at 2 consecutive days into day 1–3 neonatal mice. The first injection was always performed at the day of T cell transfer.

For the in vitro stimulation of CD8⁺Thy1.1⁺ OT-I T cells, spleen cells from the respective recipient mice were cultured individually in 96-well tissue culture plates for 4–6 h at 37°C in complete RPMI (RPMI 1640 with 10% FCS, penicillin/streptomycin, 2-ME [50 μM], and nonessential amino acids) with or without 10 μM peptide OVA_257–264 (SIINFEKL) in the presence of brefeldin A. For intracellular IFN-γ staining, an intracellular staining kit from BD Pharmingen (Hamburg, Germany) was used. The samples were analyzed with a FACSCalibur flow cytometer (Becton Dickinson). Specificity of staining was confirmed using isotype-matched control Abs.

To increase dendritic cell (DC) numbers, RAG^−/− mice were injected with B16 melanoma cells secreting GM-CSF. DCs were purified from spleens of tumor-bearing RAG^−/− mice using CD11c-specific magnetic beads and autoMACS (both Miltenyi Biotec). The DC preparations contained 80% CD11c⁺ cells as judged by flow cytometry using a CD11c-specific mAb. DCs were washed twice in PBS, aliquoted, and stored in 90% FCS and 10% DMSO at −70°C. For vaccination, DCs were thawed, washed twice in complete RPMI, and incubated with 10 μM peptide OVA_257–264 or peptide βGal_497–504 (ICPMYARV) for 4 h at 37°C. Prior to injection, the cells were washed twice with PBS.

mAbs for flow cytometry included CD4 (RM4-5), CD8α (53-6.7), CD3ε (500A2), CD11c (HL3), Thy1.1 (OX-7), Thy1.2 (30H12), CD44 (IM7), IFN-γ (XMG1.2), CD45 (30-F11), α₄β₇ (DATK32) (all BD Pharmingen), and IL-7Rα (A7R34; eBioscience, San Diego, CA).

From our previous work we know that lymphopenia in B6 neonates is limited to the first week of life. During this period, a small fraction of adoptively transferred naive CD44^low polyclonal CD8⁺ T cells undergo LIP in B6 neonates. Importantly, however, most naive CD8⁺ T cells do not undergo LIP (figure 3A in Ref. 10). To test whether neonatal non–T cells suppress LIP, naive TCR-transgenic (TCR^tg) CD8⁺Thy1.1⁺ OT-I T cells were labeled with CFSE and transferred into Thy1.2⁺ T cell–deficient RAG^−/− mice. As expected, OT-I cells divided in adult RAG^−/− mice (Fig. 1A, 1G). In contrast, LIP was strongly suppressed in neonatal RAG^−/− mice (Fig. 1B, 1G). ILCs modulate T cell responses, express Thy1.2, and can be depleted by the Thy1.2-specific Ab 30H12 (13–18). When neonatal RAG^−/− mice were treated with 30H12, proliferation of Thy1.1⁺ OT-I cells was restored (Fig. 1C, 1G), indicating that neonatal Thy1.2⁺ innate immune cells supppressed LIP. Next, T cell–competent Thy1.2⁺ B6 mice were reconstituted with CFSE-labeled Thy1.1⁺ OT-I cells. As expected, OT-I cells did not undergo LIP in adult B6 mice (Fig. 1D, 1H). However, the frequency of proliferating OT-I cells was significantly increased in B6 neonates (Fig. 1E, 1H) and reached values (∼30%) similar to those observed previously for polyclonal CD8⁺ T cells (figure 3A in Ref. 10). In agreement with our observation in RAG^−/− neonates (Fig. 1C, 1G), the depletion of Thy1.2⁺ cells further promoted LIP of OT-I cells in B6 neonates (Fig. 1F, 1H). Hence, the data shown so far strongly suggest that Thy1.2⁺ innate immune cells modulate the degree of LIP in neonatal mice.

This T cell–independent suppression of LIP was confirmed with polyclonal CD44^lowCD8⁺Thy1.1⁺ T cells. As shown in Fig. 1I–K, 25% of naive CD8⁺ T cells underwent LIP in RAG^−/− neonates whereas the rest did not proliferate. In apparent contrast, LIP of CD8⁺ T cells was very efficient in RAG^−/− adults (Fig. 1L–N). Thus, neonatal Thy1.2⁺ innate immune cells appear to block LIP of both TCR^tg and polyclonal CD8⁺ T cells. Importantly, however, note that still some TCR^tg (Fig. 1G, 1H) and polyclonal CD8⁺ T cells (Fig. 1I–K) underwent LIP in neonates and escaped suppression by Thy1.2⁺ innate immune cells. The molecular basis for this escape remains to be defined. Nevertheless, it would explain why some LIP-dependent CD44^high memory T cells can be generated in neonatal mice (8–10) whereas LIP of many T cells is suppressed by Thy1.2⁺ innate immune cells at the same time (Fig. 1).

Based on the expression of subset-specific transcription factors, ILCs can be subdivided into T-bet⁺ ILC1s, GATA3⁺ ILC2s, and RORγt⁺ ILC3s (19, 20). ILC development and/or function are severly impaired in the absence of IL-7 or its receptor (IL-7R) (21–24) (data not shown). Given that ILCs block LIP in neonates, spontaneous CD8⁺ T cell activation and LIP should occur in IL-7R–deficient (IL-7R^−/−) neonates. This was indeed the case. The few T cells found in spleens of 10-d-old IL-7R^−/− neonates mainly displayed an activated phenotype as shown by their high CD44 and low CD62L expression (Fig. 2A). To test whether the IL-7R^−/− environment was sufficient for CD8⁺ T cell activation, CFSE-labeled CD8⁺ OT-I T cells were transferred into neonatal IL-7R^−/− mice. There, the degree of LIP (Fig. 2B) was comparable to that observed in anti-Thy1.2–treated RAG^−/− and B6 neonates (Fig. 1C, 1F). Next, JAK3-deficient (JAK3^−/−) neonates were reconstituted with OT-I cells. In these mice, the IL-7R is still expressed by the remaining ILCs (data not shown) but IL-7R signaling is defective. Similar to IL-7Rα^−/− neonates, OT-I cells underwent LIP in JAK3^−/− mice (Fig. 2B). Taken together, the data presented so far suggest that IL-7R signaling in Thy1.2⁺ innate immune cells contributes to the suppression of CD8⁺ T cell LIP in neonates.

ILC3s can present Ag via MHC class II and modulate CD4⁺ T cell responses (15, 17). Lymphoid tissue-inducing (LTi) cells belong to the ILC3 subset and express both Thy1.2 and IL-7R (19). Furthermore, they can be found in spleen (25) and neonatal lymph nodes (26) where LIP is initiated (27). Importantly, CD4⁺ LTi cells are particularly abundant in neonatal mice and their frequency declines with increasing age (28). Accordingly, we detected considerable numbers of CD3⁻ lymphoid cells in spleens of neonatal B6 and RAG^−/− mice, which expressed CD4, IL-7R (Fig. 3A, 3B), Thy1.2, α₄β₇, and the RORγt (data not shown), which is crucial for LTi cell development (29). Importantly, CD4⁺ LTi cells were senstitive to anti-CD4–mediated depletion (Fig. 3C) (30).

To determine whether CD4⁺ LTi cells are sufficient for the suppression of LIP, RAG^−/− neonates were treated with anti-CD4 Abs and reconstituted with either CFSE-labeled OT-I cells or polyclonal CD44^lowThy1.1⁺CD8⁺ T cells. Anti-CD4 treatment enabled LIP of both CD8⁺ T cell populations (Fig. 3E, 3F). This was not the case in PBS-treated controls (Fig. 3D).

Next we studied whether anti-CD4–induced LIP leads to the functional maturation of OT-I cells in neonates. For this purpose, CFSE-labeled OT-I cells were transferred into anti-CD4–treated RAG^−/− neonates. After 10–14 d, recipient spleens were cultured for 4–6 h in vitro in the presence or absence of the OT-I–specific peptide OVA_257–264. Most OT-I cells had divided and 35% of the cells produced IFN-γ in the presence of their cognate peptide (Fig. 3G) but not in its absence (Fig. 3H). Furthermore, only OT-I cells that had divided two or more times produced considerable amounts of IFN-γ, which increased proportionally with the number of cell divisions.

Importantly, neonatal CD4⁺ LTi cells did not block Ag-specific priming. When untreated RAG^−/− neonates were reconstituted with CFSE-labeled OT-I cells and immunized with OVA_257–264-loaded DCs, OT-I cells proliferated and differentiated into IFN-γ–producing effector/memory T cells (Fig. 3I). This was not the case after vaccination with control peptide-loaded DCs (Fig. 3J). Thus, CD4⁺ innate immune cells suppress LIP and subsequent IFN-γ production but not Ag-specific priming of CD8⁺ T cells.

ILCs consist of multiple cell types that modulate tissue homeostasis and immune responses (19, 20). Recently, it was shown that ILC3s can present MHC class II–restricted Ags and counterregulate CD4⁺ T cell responses (15, 17). Whether the inhibition of CD4⁺ T cell LIP in neonates (Supplemental Fig. 1) relies on direct ILC3–CD4 interactions remains open. However, to our knowledge, the modulation of CD8⁺ T cell homeoastsis by ILCs has not been described before.

Although we cannot fully exclude the contribution of other innate immune cells, our data strongly suggest that neonatal CD4⁺Thy1.2⁺ ILC3s, most probably LTi cells, prevent LIP of CD8⁺ T cells in neonates. At the time of birth, CD4⁺Thy1.2⁺ LTi cells are particularly frequent and their number declines with age (26, 28). An age-related loss of suppressive ILC subsets may explain why CD8⁺ T cell LIP is only moderately increased in 30H12-treated adult RAG^−/− mice (Supplemental Fig. 2). Alternatively, neonatal ILCs might lose their suppressive function in an age-dependent manner. Nevertheless, suppression of LIP requires host IL-7R signaling in neonates. Whether IL-7R–dependent suppression of LIP involves other cells than LTi cells remains open.

In summary, we show that T cell homeostasis in neonatal mice is regulated by mechanisms that are fundamentally different to adults. Our data help to explain how immunological self-tolerance can be established in lymphopenic neonates (31) whereas lymphopenia in the adult causes autoimmunity (3).

We thank A. Frenznik, M. Jaster, Gorana Hollmann, Carmen Henrich-Kellner, N. Michel, and S. Prokosch for excellent technical assistance and N. Garbi, T. Kammertöns, Mario Berger, and S. Krauß-Schüler for critical discussions.

The authors have no financial conflicts of interest.