Costimulation blockade in combination with IL-2 permits regulatory T cell sparing immunomodulation that inhibits autoimmunity

Abstract

Blockade of CD28 costimulation with CTLA-4-Ig/Abatacept is used to dampen effector T cell responses in autoimmune and transplantation settings. However, a significant drawback of this approach is impaired regulatory T cell homeostasis that requires CD28 signaling. Therefore, strategies that restrict the effects of costimulation blockade to effector T cells would be advantageous. Here we probe the relative roles of CD28 and IL-2 in maintaining Treg. We find provision of IL-2 counteracts the regulatory T cell loss induced by costimulation blockade while minimally affecting the conventional T cell compartment. These data suggest that combining costimulation blockade with IL-2 treatment may selectively impair effector T cell responses while maintaining regulatory T cells. Using a mouse model of autoimmune diabetes, we show combined therapy supports regulatory T cell homeostasis and protects from disease. These findings are recapitulated in humanised mice using clinically relevant reagents and provide an exemplar for rational use of a second immunotherapy to offset known limitations of the first.

Fig. 1. Impact of CD28 costimulation blockade on Treg homeostasis.

6–9 week old BALB/c mice were injected i.p. with anti-CD80/86 Ab or control Ab on d0 and d6. At d8, spleen (Spl) and peripheral lymph node (LN) cells were harvested for analysis. a Representative FACS plots show Treg percentage in gated CD4 + cells from either spleen or peripheral lymph nodes b. Collated data for Treg percentage (left) and Treg absolute number (right) (n = 7 per group). Representative FACS plots from spleen c and pooled data d illustrate Ki67 expression on Treg (CD4 + Foxp3+) or Tconv (CD4 + Foxp3-) (n = 7 per group). Data are collated from 3 independent experiments. Graphs b and d show mean±s.d.; each dot represents one mouse. **P < 0.01, ***P < 0.001, ****P < 0.0001 (two-tailed t test). Source data are provided as a Source Data file.

Fig. 2. IL-2 promotes Treg homeostasis in CD28-deficient mice.

6–8 week old BALB/c and CD28^−/− mice were injected i.p. with IL-2 complex or control on d0, d2, d4, d6 and d7. At d8, spleen cells were harvested for analysis. Graphs show collated data for Treg percentage in CD4 + cells (left), Treg absolute number (middle) and Ki67 expression on Treg or Tconv (right) from treated BALB/c a, or CD28^−/− b mice. Data are presented as mean±s.d.; each dot indicates one mouse. a n = 11 for control, n = 6 for IL-2; (b) n = 9 for control, n = 7 for IL-2. Data are collated from 5 independent experiments. *P < 0.05, **P < 0.01, ***P < 0.001, ns not significant (two-tailed t test). Source data are provided as a Source Data file.

Fig. 3. IL-2 can compensate for CD28 blockade in Treg homeostasis.

7–15 week old BALB/c mice were injected i.p. with control Ab, anti-CD80/86 Ab, IL-2 complex or both anti-CD80/86 Ab and IL-2 complex as indicated (anti-CD80/86 Ab was injected on d0 and d6; IL-2 complex was injected on d0, d2, d5 and d7). Spleen cells were analysed at d8. Graphs show collated data for Treg percentage in gated CD4 + CD3 + cells a, Treg Foxp3 MFI b, Treg absolute number c, and Tconv absolute number d. Representative FACS plots e, and pooled data depict Treg expression of CD25 f, CTLA-4 g, and TGFβ h. Graphs show mean±s.d.; each dot indicates one mouse. a, b, f, g n = 17 for aCD80/86, n = 18 for other groups; c, d n = 15 for IL-2, n = 14 for other groups; h n = 16 for control, n = 15 for aCD80/86, n = 17 for aCD80/86+IL-2, n = 15 for IL-2. Data are collated from 5 independent experiments. **P < 0.01, ***P < 0.001, ****P < 0.0001, ns not significant (ANOVA). Source data are provided as a Source Data file.

Fig. 4. Impact of costimulation blockade and IL-2 on diabetes development in a preventative setting.

a DO11 × RIPmOVA mice were treated i.p. with control Ab, anti-CD80/86 Ab, IL-2 complex or both anti-CD80/86 Ab and IL-2 complex (two doses of anti-CD80/86 and seven doses of IL-2 complex). Spleen cells were analysed at d8. Graphs show collated data for Treg percentage in gated CD4 + cells (left), Treg absolute number (middle) and Tconv absolute number (right). Data are presented as mean±s.d.; each dot indicates one mouse. n = 8 for control, n = 4 for aCD80/86, n = 5 for aCD80/86+IL-2, n = 4 for IL-2. Data are collated from 3 independent experiments. **P < 0.01, ****P < 0.0001, ns = not significant (ANOVA). b 2-phase combination treatment protocol. 4–6 week old normoglycaemic DO11 × RIPmOVA mice were treated with anti-CD80/86 Ab plus IL-2 complex for 7 weeks (phase 1). Subsequently, mice were maintained on IL-2 complex alone (phase 2). Control groups received either treatment alone or control Ab (not depicted). c Percentage of non-diabetic mice based on blood glucose measurements (n = 10 for control, n = 6 for anti-CD80/86, n = 9 for anti-CD80/86+IL-2, n = 7 for IL-2; Data are collated from 3 independent experiments). P values were determined by Log-rank test with Bonferroni correction. Values for the anti-CD80/86 Ab group (**P = 0.0017) and anti-CD80/86 Ab plus IL-2 complex group (****P < 0.0001) were significantly different from the control group. d At the end of the experiment, frozen pancreas sections from 6 mice for each treatment group were stained for CD4 and insulin by immunohistochemistry and the percentage of pancreas area positive for insulin was calculated for each mouse (n = 6 per group, total of 182 islets for control group, 292 islets for anti-CD80/86 group, 424 islets for anti-CD80/86+IL-2 group, 296 islets for IL-2 group). ***P < 0.001, ns not significant (ANOVA). e Frozen pancreas sections from mice from each treatment group were stained by immunofluorescence for the indicated markers. Representative confocal microscopy images are shown (20× magnification, 1.75 zoom). Source data are provided as a Source Data file.

Fig. 5. Impact of costimulation blockade and IL-2 on diabetes development in a therapeutic setting.

a Schematic of treatment protocol. Blood glucose levels of DO11 × RIPmOVA mice were tracked, and animals with values falling between 180 mg/dL and 290 mg/dL were identified for treatment (see also Supplementary Fig. 4). Mice were injected i.p. with control Ab, anti-CD80/86 Ab, IL-2 complex or both anti-CD80/86 Ab and IL-2 complex. Anti-CD80/86 Ab was given 2 doses per week for total 10 weeks; IL-2 complex was administered in 3 doses for the first week and then 2 doses per week. b Percentage of non-diabetic mice based on blood glucose measurements; n = 17 for control, n = 8 for aCD80/86, n = 10 for aCD80/86+IL-2, n = 8 for IL-2. Data are collated from 3 independent experiments. Diabetes incidence in the anti-CD80/86 Ab plus IL-2 complex group, but not other treatment groups, was significantly different from the control group (P < 0.01, Log-rank test with Bonferroni correction). c Collated data showing the percentages of DO11+ Treg in peripheral lymph nodes (LN), pancreatic lymph nodes (PanLN), spleen and pancreas. d Collated data showing CD25 and CTLA-4 expression on gated DO11+ Tconv or e DO11+ Treg in the pancreatic lymph nodes. Data are presented as mean ± sd; each dot indicates one mouse. c–e n = 10 for control, n = 5 for aCD80/86, n = 6 for aCD80/86+IL-2, n = 4 for IL-2. Data are collated from 3 independent experiments. **P < 0.01, ***P < 0.001, ****P < 0.0001, ns not significant (ANOVA). Source data are provided as a Source Data file.

Fig. 6. IL-2 restores Treg homeostasis after costimulation blockade in humanised mice.

4–6 week old, irradiated NSG mice (0.8 Gy) were adoptively transferred with 2 × 10⁵ CD34 + cells isolated from human cord blood. 16–23 weeks later, reconstituted humanised mice were treated with control Ab, Abatacept, IgG-(IL-2)₂ or both Abatacept and IgG-(IL-2)₂ as indicated. Abatacept was injected i.p. on d0, d3 and d6; IgG-(IL-2)₂ was injected s.c. on d0, d4 and d6. IgG-(IL-2)₂ is depicted as IL-2 in the figure for clarity. At d7, spleen cells were harvested for analysis. Graphs show collated data for the percentage Treg in gated CD4 + CD3 + cells a, Treg Foxp3 MFI b, Treg absolute number c, and Tconv absolute number d. Representative FACS plots e, and pooled data depict Treg expression of CD25 f, CTLA-4 g, and TGFβ h for each treatment group. Graphed data are presented as mean±sd; each dot indicates one mouse. (a–d and f, g) n = 12 for control, n = 14 for Abatacept, n = 17 for Abatacept+IL-2 and IL-2 alone; Data are collated from 5 independent experiments. h n = 9 for control, n = 10 for Abatacept, n = 7 for Abatacept+IL-2 and IL-2 alone. Data are collated from 3 independent experiments. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001, ns not significant (ANOVA). Source data are provided as a Source Data file.