Role of conserved non-coding DNA elements in the Foxp3 gene in regulatory T-cell fate

Abstract

Immune homeostasis is dependent on tight control over the size of a population of regulatory T (T(reg)) cells capable of suppressing over-exuberant immune responses. The T(reg) cell subset is comprised of cells that commit to the T(reg) lineage by upregulating the transcription factor Foxp3 either in the thymus (tT(reg)) or in the periphery (iT(reg)). Considering a central role for Foxp3 in T(reg) cell differentiation and function, we proposed that conserved non-coding DNA sequence (CNS) elements at the Foxp3 locus encode information defining the size, composition and stability of the T(reg) cell population. Here we describe the function of three Foxp3 CNS elements (CNS1-3) in T(reg) cell fate determination in mice. The pioneer element CNS3, which acts to potently increase the frequency of T(reg) cells generated in the thymus and the periphery, binds c-Rel in in vitro assays. In contrast, CNS1, which contains a TGF-beta-NFAT response element, is superfluous for tT(reg) cell differentiation, but has a prominent role in iT(reg) cell generation in gut-associated lymphoid tissues. CNS2, although dispensable for Foxp3 induction, is required for Foxp3 expression in the progeny of dividing T(reg) cells. Foxp3 binds to CNS2 in a Cbf-beta-Runx1 and CpG DNA demethylation-dependent manner, suggesting that Foxp3 recruitment to this 'cellular memory module' facilitates the heritable maintenance of the active state of the Foxp3 locus and, therefore, T(reg) lineage stability. Together, our studies demonstrate that the composition, size and maintenance of the T(reg) cell population are controlled by Foxp3 CNS elements engaged in response to distinct cell-extrinsic or -intrinsic cues.

Figure 1. Conserved non-coding sequences and chromatin modifications at the Foxp3 locus

a, Comparison of mouse Foxp3 genomic sequence to human, dog and rat. b–e, Map of permissive chromatin modifications at the Foxp3 locus by ChIP–qPCR with primers spaced at ∼1-kb intervals for B220⁺ B cells (B), CD4⁺CD25⁻ T_N cells, and CD4⁺CD25⁺ T_reg (T_R) cells. Relative (rel.) enrichment data are shown for H3K9/14Ac (b), H3K4me3 (c), H3K4me2 (d) and H3K4me1 (e). f, H3K4me1 ChIP–qPCR as in e, for CD4⁺CD8⁺ (double positive, DP) and CD4⁺CD8⁻ (CD4 single positive, CD4SP) thymocytes, and T_reg cells. In b–f, green bars denote the promoter (Pro) and CNS1–3. g, NF-κB consensus motif in CNS3 (core motif in bold) and homologous CD28RE motifs from Il2 and Fas. N, any base; R, purine; Y, pyrimidine. h, Binding of NF-κB family member c-Rel but not p65 to the CD28RE-like element at CNS3. Nuclear lysates from unstimulated or stimulated (1 μg ml⁻¹ CD3 and CD28 antibodies, 2 ng ml⁻¹ TGF-β) T_N cells were incubated with biotinylated double-stranded (ds)DNA probes containing the full-length CNS3 sequence (CNS3 WT), the full-length CNS3 sequence with the mutated core c-Rel motif (CNS3 Δc-Rel), the core CNS3 c-Rel motif, Il2 CD28RE (positive control), and Il2 ΔCD28RE (negative control). dsDNA probes and bound protein were precipitated by streptavidin beads and subjected to c-Rel and p65 western blot analysis.

Figure 2. CNS3 controls de novo Foxp3 expression

a, Frequency of Foxp3⁺ T_reg cells among CD4⁺CD8⁻ cells from thymus and spleen of 2-week-old CNS3-KO mice or littermate controls. b, Frequency of Ki67⁺ dividing thymic or splenic Foxp3⁺ T_reg cells from same mice as in a. c, d, Analysis of c-Rel-KO (Ly5.2) and wild-type (WT; Ly5.1) mixed bone marrow chimaeras and CNS3-KO (Ly5.2) and wild-type (Ly5.1) mixed bone marrow chimaeras. c, Frequency of Foxp3 expressing cells among CD4SP thymocytes of wild-type (Ly5.1⁺) or c-Rel-KO (Ly5.2⁺) origin. d, Ratio of Ly5.2⁺ (wild-type and c-Rel-KO, left, or wild-type and CNS3-KO, right) cells to Ly5.1⁺ (wild-type) cells within thymic CD4SP Foxp3⁺ population. Error bars (a, b, d) denote mean ± s.d.

Figure 3. CNS1 controls peripheral, but not thymic, induction of Foxp3 expression

a, Frequency of Foxp3⁺ T_reg cells among thymic CD4SP cells in CNS1-KO mice or littermate controls. NS, not significant. b, Frequency of Foxp3⁺ iT_reg cells generated after stimulation of T_N cells from CNS1-KO or littermate control mice with anti-CD3 (1 μg ml⁻¹), TGF-β (2 ng ml⁻¹) and Ly5.1⁺ irradiated (20 Gy) T-cell-depleted splenocytes for 72h. c, Frequency of in vivo generated Foxp3⁺ iT_reg cells among CD4⁺ cells in spleen, Peyer's patches, intraepithelial lymphocytes (IELs), and lamina propria lymphocytes (LPLs). CNS1-KO or wild-type Ly5.2⁺CD4⁺Foxp3⁻ naive T cells were co-transferred with wild-type Ly5.1⁺Foxp3⁺ T_reg cells into T-cell-deficient recipient mice. Ten weeks later, Ly5.2⁺Foxp3⁺ cell populations were analysed by flow cytometry. Error bars denote mean ± s.d. d, Frequency of Foxp3⁺ T_reg cells among CD4⁺ T cells in peripheral lymph nodes (LN; ‘non-mesenteric’), and Peyer's patches of 8–11-month-old CNS1-KO or littermate control mice.

Figure 4. CNS2 controls the heritable maintenance of Foxp3 expression

a, Frequency of Foxp3⁺ T_reg cells among CD4SP thymocytes or splenic and lymph node CD4⁺ cells in 6-month-old CNS2-KO mice or littermate controls. b, Maintenance of Foxp3 expression in T_reg cells from Ly5.2⁺ CNS2-KO or wild-type mice 4 weeks after co-transfer with wild-type Ly5.1⁺CD4⁺Foxp3⁻ cells into T-cell-deficient recipient mice. c, Cell division (div.) results in lower Foxp3 expression level and frequency in CNS2-KO T_reg cells compared to wild-type T_reg cells. CFSE-labelled CNS2-KO or wild-type Foxp3⁺ T_reg cells were cultured in vitro for 3 days before flow cytometric analysis of Foxp3 expression. MFI, mean fluorescence intensity. d, Foxp3 binds to dsDNA probes containing CNS2 sequences. Biotinylated dsDNA probes containing CNS1–3 sequences were incubated with nuclear extracts from T_N or T_R cells, precipitated with streptavidin beads and subjected to Foxp3 western blot analysis. e, Anti-Foxp3 ChIP was performed using wild-type CD4⁺CD25⁺ T_reg cells to detect Foxp3 binding to CNS regions in vivo. The Foxp3 target gene Ikzf2 was used as a positive control, Gmpr as a non-Foxp3-binding negative control. f, Percentages of CpG dinucleotide methylation at CNS2 determined by bisulphite pyrosequencing in T_N cells, T_N cells stimulated for 72 h with anti-CD3 and anti-CD28 antibodies (stim. T_N), purified Foxp3⁺ induced T_reg (iT_R, T_N stimulated as above with 2 ng ml⁻¹ TGF-β), CD4⁺CD25⁺ T_reg (T_R) cells, and T_FN from Foxp3^gfpko mice. g, Foxp3 ChIP at CNS2, the Foxp3 target gene Pde3b (positive control), and Gmpr (negative control) in ex vivo T_reg cells and iT_reg cells collected on days (D) 5 and 7 of culture with 1μg ml⁻¹ plate-bound CD3 and CD28 antibodies and 2 ng ml⁻¹ TGF-β. h, CpG methylation-sensitive-binding of Foxp3, Runx1 and Cbf-β was determined using methylated and demethylated biotinylated dsDNA CNS2 probes incubated with nuclear extracts from T_N and T_R before streptavidin precipitation and western blot analysis as in d. i, High-resolution Foxp3 ChIP–qPCR at CNS2 (for primer sequences see Supplementary Table 2) (left) and Foxp3 ChIP in wild-type and Cbf-β-deficient T_reg cells from Cbfb^fl/flFoxp3^Cre mice (right). Green bar denotes the CNS2 region. Foxp3 binding to Nt5e and Ikzf2 served as positive and to Gmpr as negative controls. Error bars (c, e–g, i) denote mean ± s.d.