CTLA-4 Controls Regulatory T Cell Peripheral Homeostasis and Is Required for Suppression of Pancreatic Islet Autoimmunity

Foxp3⁺ cells in the periphery of CTLA-4-deficient mice show increased proliferation. A, Spleen sections from 17-day-old CTLA4^−/− mice or littermate controls (CTLA4^+/−) were stained for CD4 (blue), Foxp3 (red), and Ki67 (green). Proliferating Treg appear yellow. Lower panel shows scoring of stained sections for the percentage of Foxp3⁺ cells that are Ki67⁺. Each symbol represents a different field of view and each column a different mouse. In brief, ≥5 fields were scored per mouse. Mean values were significantly different (p < 0.05). B, Spleen cells from 17-day-old CTLA4^−/− mice or littermate controls (CTLA4^+/−) were stained for CD4 and intracellular Foxp3 and Ki67. Plots are gated on CD4⁺Foxp3⁺ cells. Treg proliferation in age-matched wild-type mice (CTLA-4^+/+) was similar to that seen in CTLA-4^+/− mice. Data are representative of two (A) or more than five (B) independent experiments.

Analysis of Ag-specific Treg deficient in CTLA-4. Treg were quantified by Foxp3 staining in DO11 × RIP-mOVA/rag^−/− mice sufficient or deficient for CTLA-4. Ag-negative DO11 mice were used as controls. All mice were rag gene deficient. A, Representative Foxp3 staining in gated CD4⁺ (spleen) or CD4⁺CD8⁻ (thymus (Thy)) cells from 3-wk-old animals. B, Percentage of CD4 cells (CD4⁺CD8⁻ in thymus) that express Foxp3 at the indicated time after birth in DO11 × RIP-mOVA/rag^−/− or DO11 × RIP-mOVA/ctla4^−/−rag^−/− mice. Each symbol represents an individual mouse. C, Absolute number of CD4⁺CD8⁻Foxp3⁺ (thymus) or CD4⁺Foxp3⁺ cells (spleen) in 6-wk-old mice of the indicated genotype. D, Pooled pancreata from DO11 × RIP-mOVA/rag^−/− or DO11 × RIP-mOVA/ctla4^−/−rag^−/− mice were digested and stained for CD4, Foxp3, and Ki67. Plots are gated on CD4⁺ cells and are representative of two similar experiments using pooled pancreas tissue from two to three animals.

Anti-CTLA-4 Ab increases Treg proliferation in vivo. A, BALB/c mice were injected with 500 μg of anti-CTLA-4 Ab (or control hamster IgG) on days 0, 2, and 6 and LN were harvested on day 8. Cells were stained for CD4 and intracellular Foxp3 and Ki67. Representative FACS stains showing Ki67 expression in BALB/c LN cells gated on CD4⁺Foxp3⁺ or CD4⁺Foxp3⁻ cells. Proliferation of CD4⁺Foxp3⁺ cells was increased in mice treated with anti-CTLA-4 Ab compared with control Ab (30.68 ± 0.9% vs 13.85 ± 2.6%, respectively; n = 4). B, BALB/c mice were treated with twice weekly injections of anti-CTLA-4 Ab or control Ab and analyzed at the indicated time points. The percentage of Ki67⁺ cells within the CD4⁺Foxp3⁺ (Treg) population and the CD4⁺Foxp3⁻ (T cell) population in LN is shown. Each bar shows the mean value for three to six individual mice with SD (∗∗, p < 0.005). Values from control Ab-treated mice were similar at all time points and are presented as a single pooled control (individual controls were used for statistical analysis). C, CD69 staining in gated CD4⁺Foxp3⁺ Treg and CD4⁺Foxp3⁻ T cells 24 h following injection of BALB/c mice with 500 μg of anti-CTLA-4 Ab or control Ab. D, Mean values with SD for multiple mice treated as in C (∗, p < 0.01; n = 4).

Treg deficient in CTLA-4 are not inherently pathogenic. One to 2 × 10⁶ LN cells or purified CD4⁺CD25⁻ or CD4⁺CD25^high LN cells from 15- to 17-day-old CTLA-4^−/− mice were injected into rag^−/− recipients. A, Percentage weight loss and LN cellularity are shown 3–4 wk following transfer. Columns show mean data with SD for three to five mice. B, H&E staining of heart from rag^−/− mice that received CTLA-4^−/− whole LN, CD4⁺CD25⁻ cells, or CD4⁺CD25^high cells.

Ag-specific Treg deficient in CTLA-4 lack regulatory function in vivo. A, Foxp3 staining in DO11⁺CD4⁺CD25⁺ LN cells from DO11⁺mOVA⁺ rag^−/− or DO11⁺mOVA⁺ ctla4^−/−rag^−/− mice. B, Blood glucose levels in mOVA/rag mice that received 0.15 × 10⁶ DO11⁺CD25⁻ cells alone or in combination with 0.15 × 10⁶ CD4⁺CD25⁺ cells purified from DO11⁺mOVA⁺ rag^−/− or DO11⁺mOVA⁺ ctla4^−/−rag^−/− mice. Six mice are shown per experimental group, except for the CD25⁻ alone group where only two are shown for clarity. Diabetes was observed for zero of six recipients of wild-type (wt) Treg, six of six recipients of CTLA-4^−/− Treg, and six of six recipients of CD25⁻ cells alone. Transfer of CTLA-4^−/− Treg alone did not induce diabetes. C, Digested pancreas samples from recipients of wild-type or CTLA-4^−/− DO11⁺ Treg 22 days after transfer were stained for CD4 and Foxp3. Absolute cell numbers were calculated as the mean value for three recipients: for CD4⁺Foxp3⁻ cells, these were 1648 and 4719 (for recipients of wild-type and ctla4^−/− Treg, respectively) and for CD4⁺Foxp3⁺ cells these were 335 and 440 (for recipients of wild-type and ctla4^−/− Treg, respectively). D, 2.5 × 10⁴ Thy1.1⁺CD4⁺CD25⁻ cells were incubated with wild-type or ctla4^−/− CD4⁺CD25⁺ cells (upper panel) or control CD25⁻ cells (lower panel) in the presence of 1 μg/ml anti-CD3 and APC. Seventy-two hours later, the absolute number of Thy1.1⁺ responder cells was assessed by flow cytometry. Results are expressed as a percentage of the maximum cell number and graphs show the mean value and SD from three independent experiments. E, LN cells from wild-type or ctla4^−/− DO11⁺mOVA⁺ rag^−/− mice were stained for surface TGFβ and secreted IL-10 as described in Materials and Methods. Plots are gated on CD4⁺Foxp3⁺ (Treg) or CD4⁺Foxp3⁻ (T cells).