A pegylated IL-2 mutein biased for CD25hi T cell engagement specifically promotes FoxP3+ Treg expansion in vitro. The majority of FoxP3+ Tregs constitutively express CD25 (IL-2Rα) at relatively high levels, making them particularly sensitive to low concentrations of IL-2. However, CD8+ T cells and NK cells readily respond to IL-2 by expressing high amounts of the lower affinity CD122 (IL-2Rβ) chain and, as such, represent a major hurdle in the development of IL-2–modulating therapies aimed at promoting Treg-mediated immune tolerance (35). To investigate the target cell selectivity and potency of SAR’336, we exposed (CD3+CD4+FoxP3+) Tregs, (CD3+CD8+) CD8+ T cells, and (NK1.1+CD3–) NK cells isolated from the spleens of C57BL/6 mice to the mutein. Recombinant human IL-2 (rhIL-2) potently stimulated the phosphorylation of intracellular STAT5 in murine FoxP3+ Tregs and at higher concentrations in CD8+ T cells and NK cells (Figure 1A), verifying the higher sensitivity of Tregs to IL-2. On the other hand, SAR’336 specifically induced the phosphorylation of STAT5 in FoxP3+ Tregs, not NK or CD8+ T cells (Figure 1B). To verify that the signal SAR’336 provided depended on CD25 expression at the cell surface, we repeated the assay and segregated FoxP3+ Tregs based on the level of CD25 expression at the time of p-STAT5 staining (Figure 1C). SAR’336 preferentially induced the phosphorylation of STAT5 in CD25hi Tregs, consistent with the selective nature of the mutein (Figure 1D). Thus, SAR’336, as designed (27), specifically targets CD25hi Tregs, while rhIL-2 activates both CD25– and CD25+ Tregs.

A pegylated IL-2 mutein targeting the CD25/STAT5 signaling pathway promotes specific Tregs’ expansion and Treg-associated gene expression. (A–C) Geometric mean fluorescence intensity (MFI) of phosphorylated STAT5 (pSTAT5) in FoxP3+CD4+CD3+ T cells (Tregs), NK1.1+CD3– cells (NK cells), and CD8+CD3+ T cells (CD8+ T cells) isolated from C57BL/6 mice and exposed to increasing concentrations of rhIL-2 (A) or SAR’336 (B) for 45 minutes. Data representative of more than 1 study. (C) Representative flow cytometry of the gating strategy for the identification of murine CD25hi and CD25lo FoxP3+ Tregs at 45 minutes. (D) Effect of the concentration (pg/mL) of rhIL-2 or SAR’336 on the geometric MFI of p-STAT5 expression in total FoxP3+, CD25hiFoxP3+, and CD25loFoxP3+ Tregs. The red line represents the average of FoxP3+ T cells in the presence of rhIL-2, and the green line represents the average of FoxP3+ T cells in the presence of SAR’336. (E–H) Murine CD4+GFP+ Tregs from the spleens of B6.FoxP3GFPki mice were purified and activated by plated α-CD3 and α-CD28 for 72 hours in the presence of rhIL-2 or SAR’336. (E) Representative histogram of the expression of Ki-67 at 72 hours. (F) Effect of increasing dose of SAR’336 and rhIL-2 on the frequency of Ki-67+ among live CD4+FoxP3+ cells at 72 hours. (G) Effect of increasing doses in the total number of live Tregs at 72 hours. Two-way ANOVA. (H) Percentage increase in geometric MFI of Helios, FoxP3, and CD25 in the presence of 1 μg/mL of each cytokine/mutein over the medium alone (0 μg/mL). (Mean MFI of the experiment/mean MFI of medium alone) (n = 3 per experiment, 3 individual experiments). Two-way ANOVA. (I) CD4+GFP+ Tregs were cocultured with CellTrace Violet–labeled (CTV-labeled) CD4+GFP– Teffs in the presence of MitoC-treated antigen-presenting cells (APCs) (CD4– fraction) and soluble α-CD3 (1 μg/mL) with 1 μg/mL of rhIL-2 or SAR’336 for 72 hours. Percentage increase in the number of Tregs relative to medium: (# cells in rhIL-2 or SAR’336/mean # in medium) × 100. Compiled results of 3 distinct experiments with triplicates. One-way ANOVA. Tukey’s correction. ***P < 0.01.

Next, we addressed the capacity of SAR’336 to promote Treg fitness in vitro. While rhIL-2 caused an increase in the expression of the mitotic marker Ki-67 in Tregs (Figure 1, E and F), SAR’336 required higher doses for all Tregs present in the culture to engage in mitosis. However, both rhIL-2 and SAR’336 led to a dose-dependent increase in the number of FoxP3+ Tregs by 72 hours in culture (Figure 1G). This is consistent with the fact SAR’336 requires higher concentrations to fully engage STAT5 signaling. Nonetheless, SAR’336 induced similar levels of the transcription factor Helios, FoxP3, and CD25 expression (increase in MFI) as rhIL-2 in TCR-activated cells in vitro (Figure 1H), verifying that the molecule acted as a bona fide IL-2 signal in Tregs. Helios, a transcription factor of the Ikaros family, promotes the transcriptional stability of FoxP3 in Tregs (36, 37), notably, by supporting the IL-2/STAT5 signaling pathway (38), and these results suggest Helios+ cells are particularly sensitive to SAR’336. Finally, since conventional effector CD4+ T cells (Teffs) can upregulate the CD25 receptor upon activation (39), we cocultured activated FoxP3+ Tregs with Teffs (1:4 ratio) to mimic a competitive environment. SAR’336, but not rhIL-2, generated increased frequencies of Tregs over Teffs (Figure 1I), demonstrating that the engineered molecule maintained its selectivity for Tregs. Collectively, these results illustrate key differences in how the pegylated SAR’336 mutein provides a p-STAT5 signal to promote the proliferation and fitness of Helios+CD25hi Tregs.

SAR’336 stimulates the rapid and specific expansion of FoxP3+ Tregs in pancreatic islets. It is well established that CD4+CD25+ Tregs offer protection against diabetes in the NOD model (40–42), where they depend on IL-2 to exert their suppressive functions (43). To dissect the systemic and local effects of the SAR’336 mutein on local immune responses, we administered a single dose of SAR’336 to young, female NOD and NOD BDC2.5 (nondiabetic) mice (in which CD4+ T cells are specific for the chromogranin A autoantigen expressed by β cells) and assessed the frequency of FoxP3+CD4+ T cells in the blood (PBMCs), spleen, peripheral lymph nodes (pLNs), and pancreas at days 2 and 4 (Figure 2A). In both groups, we observed that the frequency of Tregs among CD4+ T cells was highest by day 4 of injection in blood as well as in the spleen, LNs, and pancreas (Figure 2B), consistent with the recently reported pharmacodynamics of SAR’336 in wild-type C57BL/6 mice (27). Importantly, SAR’336 increased the proportion of FoxP3+ Tregs over Teffs and NK cells while not affecting other immune cells in the pancreas (Figure 2C). Nonetheless, in contrast with events in the pancreatic tissue, the frequency of NK cells increased slightly among circulating PBMCs (Supplemental Figure 1; supplemental material available online with this article; https://doi.org/10.1172/jci.insight.182064DS1), prompting us to investigate the off-target effects of SAR’336 in the spleen and pLNs (axillary and inguinal). Here, the number of NK cells was not increased (Supplemental Figure 2, A–E) while the ratio of Tregs to NK cells or conventional T cells was higher in the spleen (Supplemental Figure 2, B–D), verifying the preferential effect of SAR’336 on the expansion of Tregs. Finally, when we stained for the mitotic marker Ki-67, we observed that, while the frequency of Ki-67+ Tregs and NK cells both increased in the spleen (Supplemental Figure 2F), there were more proliferating Tregs, rather than NK cells, in the pancreas (Supplemental Figure 2G), verifying the targeted effect of the mutein on tissue-localized Tregs.

SAR’336 promotes the rapid and specific expansion of CD4+FoxP3+ T cells. (A) Female NOD and NOD BDC2.5 mice were administered 0.3 mg/kg of SAR’336 or the vehicle (Veh) s.c., and cells from the blood (PBMCs), the spleen, inguinal and axillary LN (pLN), and the pancreas were collected at day 2 and 4 after injection. (n = 4–5/group.) (B) Frequency of FoxP3+ among CD4+ T cells isolated in each organ at days 2 and 4 after injection. (C) Pie charts representing the mean frequency of conventional CD4+ T cells (Tconv), Tregs, CD8+ T cells, and NK and B cells as parts of whole cells collected from the pancreas of NOD mice at day 4 postinjection. Two-way ANOVA. *P < 0.05; **P < 0.01; ***P < 0.001. (D and E) BDC2.5+CD4+ T cells were isolated, and 5 × 106 cells were adoptively transferred (i.v.) into NOD mice before the administration of 0.01, 0.1, and 0.3 mg/kg of SAR’336 or Veh s.c. (n = 4–5/group). (E) Number of BDC2.5 tetramer (Tet+) CD4+FoxP3– T cells in the spleen. (F) Number of CD4+FoxP3+Tet+ cells in the spleen. (G) Number of NKp46+CD3– NK cells in the spleen at day 4 after injection. One-way ANOVA. Tukey’s correction. *P < 0.05; **P < 0.01; ***P < 0.001.

Since SAR’336 efficiently promoted the expansion of Tregs in vitro, we next investigated its effect on antigen-specific T cells in NOD mice in an acute inflammatory condition in which recently activated T cells compete for IL-2 in vivo. In this context, TCR engagement also induces CD25 expression in Teffs (39), allowing them to optimize IL-2 signals during expansion, and ultimately competing with Tregs. To address the specificity of SAR’336 in an inflammatory context in vivo, BDC2.5-specific CD4+ T cells were sorted and adoptively transferred into NOD mice (44), which then received vehicle or 0.3, 0.1, or 0.01 mg/kg of SAR’336 (Figure 2D). SAR’336 expanded both antigen-specific (tetramer+; Tet+) Teffs and Tregs in the spleen at the highest dose (0.3 mg/kg; Figure 2E), yet Tregs expanded at the low and intermediate dose levels of 0.01 and 0.1 mg/kg (Figure 2F), verifying that activated, antigen-specific Tregs remain more sensitive to SAR’336. Importantly, the number of recipient NK cells remained unchanged at all concentrations (Figure 2G), verifying that SAR’336 does not promote the expansion of resting NK cells. Collectively, these results highlight the selectivity of SAR’336 to target CD25hi T cells and preferentially promote Treg expansion, even in an inflammatory setting where effector T cells upregulate CD25 surface expression.

SAR’336 minimizes cytotoxic responses and expands IL-10–producing, antigen-specific Tregs in the pancreas. Since we observed that SAR’336 induces an expansion of Tregs in the pancreas of NOD mice, we next investigated whether it prevents the development of immune cell infiltration of β islets before the onset of diabetes (hyperglycemia). To this end, we used a model of adoptive transfer of antigen-specific BDC2.5 CD4+ T cells in prediabetic NOD mice to synchronize the onset of insulitis and better trace donor cells throughout diabetes development in NOD mice (45). CD4+ T (3 × 105) cells were isolated from NOD BDC2.5 (Vβ4+) female mice and transferred i.v. into young adult female NOD mice (45) that received 0.03 mg/kg SAR’336 twice a week, an amount established previously as 1/10 of maximal dose (27) (Figure 3A). To better assess the early and late events occurring in the pancreas, pancreatic and systemic T cell populations were isolated at early (day 7) and late (day 21) stages of insulitis but before diabetes onset (Supplemental Figure 3). First, at a late time point, we observed less infiltration into the pancreatic islets in mice treated with SAR’336 relative to vehicle (Figure 3, B and C). Both the number of total cells and the number of IFN-γ–producing CD8+ T cells isolated from the pancreas at day 21 after transfer were lower in SAR’336-treated mice (Figure 3, D and E). Concomitantly, the number of IFN-γ+CD4+ T cells was reduced in the pancreas and the spleen of treated mice (Figure 3F). Administration of the engineered molecule induced an increase in the frequency of FoxP3+ Tregs among both recipient (Vβ4–) and donor (Vβ4+) CD4+ T cells in the pancreas (Figure 3, G and H) but not in the numbers of local FoxP3+ Tregs (Supplemental Figure 4A), resulting in a significantly lower ratio of IFN-γ–producing Th1 cells to Tregs in the pancreas at day 21 (Supplemental Figure 4B). Concomitantly, SAR’336 drove the accumulation of FoxP3+ Tregs in the pancreatic LN (Figure 3I), which have been shown to suppress inflammatory T cells prior to their migration to the pancreas (46, 47) by, notably, producing IL-10 (48). We observed that SAR’336 significantly increased the frequency of IL-10+ Tregs on day 21 (Figure 3J), highlighting the ability of SAR’336 to prevent the establishment of a cytotoxic immune response and promote the accumulation of IL-10+ Tregs in the pancreas.

SAR’336 minimizes cytotoxic responses and expands IL-10–producing, antigen-specific Tregs in the pancreas. (A) CD4+ T cells were isolated from female NOD BDC2.5 mice, and 3 × 105 cells were adoptively transferred i.v. into female NOD mice. A total of 0.03 mg/kg of SAR’336 was administered s.c. twice a week up to 21 days. Lymphocytes from the spleen, pancreatic LNs (panLN), and pancreas were collected at day 7 (n = 7–8/group) and day 21 (n = 3–5/group). Data compiled from 2 distinct experiments. (B) Representative histology slide of the pancreas at day 21 (hematoxylin and eosin). Black arrows show β-islet infiltration while red arrows point to perivascular infiltration of immune cells. (C) Histology score at day 21 posttransfer. Adapted from Papaccio et al. 2000 (86). 1 = infiltrates in small foci at the islet periphery; 2 = infiltrates surrounding the islets (peri-insulitis); 3 = intraislet infiltration < 50% of the islet, without islet derangement; 4 = extensive infiltration over 50% of the islet, cell destruction, and prominent cytoarchitectural derangement; 5 = complete islet atrophy and β cell loss. (D) Total number of cells isolated from the pancreas at day 7 and 21. (E) Number of IFN-γ–producing CD3+CD8β+ T cells (CD8+ T cells) in the pancreas. (F) Number of IFN-γ–producing CD3+CD4+ T cells in distinct organs at day 21 after transfer. (G) Representative flow cytometry of the expression of FoxP3 and Vβ4 in the pancreas at day 21 after transfer. (H) Frequency of FoxP3+CD4+ T cells among Vβ4+ and Vβ4– T cells at day 21 after transfer. (I) Number of FoxP3+CD4+ T cells in the pancreatic lymph node at day 7 and 21 after transfer. (J) Frequency of IL-10–producing CD4+FoxP3+ Tregs in the pancreas (Pan), pancreatic LNs (panLN), and spleen (Spl) at day 21 after transfer. One way-ANOVA. Tukey’s correction. *P < 0.05; **P < 0.01; ***P < 0.001.

SAR’336 promotes the accumulation of antigen-specific IL-10+ Tregs in β islets. Since we observed that SAR’336 promoted Treg expansion, we next investigated whether the Tregs found in the pancreas were generated locally or migrated to the site upon inflammation. While the majority of FoxP3+ Tregs found in tissues originate from thymic selection (tTregs) and migrate to the tissue, a subset of local Tregs develop from the TGF-β–dependent induction of FoxP3 (49, 50) in naive T cells, so-called peripheral Tregs (pTregs) (51). IL-2 plays a role in the accumulation of pTregs but is not directly involved in the induction of FoxP3 (52, 53), suggesting a similar mechanism for SAR’336. To verify this, CD4+FoxP3– cells isolated from B6.FoxP3GFPki reporter mice were activated in the presence of recombinant TGF-β to generate FoxP3+ T cells. SAR’336 did not further contribute to the generation of FoxP3+ T cells in vitro (Figure 4, A and B). Nonetheless, once the cells gained FoxP3 expression, SAR’336 promoted the overall level of expression of the transcription factor (MFI) (Figure 4C). To investigate whether the protective effect of SAR’336 was dependent on migrating antigen-specific Tregs, we transferred donor BDC2.5+CD4+ T cells containing Tregs (“Total CD4”) or devoid of Tregs (“Treg depleted”) into NOD recipient mice (Figure 4D and Supplemental Figure 5A). While SAR’336 promoted the generation of pTregs from donor Vβ4+CD4+ T cells in the pancreatic LN (Supplemental Figure 5B), this was not observed in the pancreas (Figure 4E). Importantly, we did not observe IL-10 production from the donor Vβ4+ pTregs when compared with the group that received total BDC2.5+CD4+ T cells (Figure 4F), verifying that the protective Tregs expanded by SAR’336 originated from the transferred antigen-specific Tregs. Collectively, these results suggest that SAR’336 promotes the expansion of migrating Tregs rather than local pTregs to control inflammation in the pancreas.

SAR’336 promotes the migration of antigen-specific, IL-10+ Tregs. (A–C) Splenic CD4+GFP– Teffs were CTV-labeled and activated in the presence of APCs and soluble anti-CD3 (1 μg/mL) with 5 ng/mL murine TGF-β and 1 μg/mL of SAR’336 for 72 hours. Representative flow cytometry of CTV and FoxP3-GFP expression. Representative of 2 distinct experiments. (B) Frequency of FoxP3+ among total live CD4+ cells. (C) Geometric MFI of FoxP3 among total CD4+ T cells. Seventy-two hours. Two-way ANOVA. ***P < 0.001. (D–F) CD4+GFP– (Treg depleted) or total CD4+ T cells were isolated from female BDC2.5 FoxP3GFPki NOD mice, and 3 × 105 cells were adoptively transferred i.v. into female NOD mice. At days 0 and 3 after transfer, 0.03 mg/kg of SAR’336 was administered s.c. One-way ANOVA. Tukey’s correction. *P < 0.05, **P < 0.01.

SAR’336 promotes GATA3 expression in expanding pancreatic Tregs. Although their transcriptional program is largely driven by the master transcription factor FoxP3 (19), Tregs acquire additional master transcription factors associated with Th cells, including GATA3, RORγT, and T-bet, to promote unique aspects of their migration, survival, and function during inflammation (20). IL-2 can directly orchestrate the transcriptional trajectory of Tregs (54, 55), notably, by promoting the generation of Tregs that express GATA3 (30) and Helios (55). Indeed, in the NOD T cell transfer experiment (Figure 3), we observed that IL-10+ Tregs SAR’336 generated (Figure 3J) in the pancreas also expressed the transcription factor Helios (Figure 5A). By day 21 after T cell transfer, SAR’336 promoted significantly more GATA3+ and less RORγT+ and T-bet+ Tregs in the pancreas relative to vehicle-treated mice (Figure 5B). Importantly, these pancreatic GATA3+ Tregs emerged from a population of ST2+ Tregs (Figure 5C), verifying a link between the expression of this transcription factor and the receptor for IL-33 (55, 56).

SAR’336 expands Helios+ST2+ Tregs in the pancreas. (A) Representative flow cytometry plot of IL-10 and Helios expression among CD4+FoxP3+ T cells in the pancreas of vehicle- (left) and SAR’336-treated (right) mice at day 21. (B) Intracellular expression of master transcription factors RORγT (red), T-bet (gray), and GATA3 (green) in CD4+FoxP3+ T cells isolated from the pancreas at day 21 after adoptive transfer. Individual Student’s t test between markers. **P < 0.01; ***P < 0.001; ###P < 0.001. (C) Representative flow cytometry plot of the expression of GATA3 and ST2 in CD3+CD4+FoxP3+ T cells isolated from the pancreas of vehicle- (veh) or SAR’336-treated mice at day 21. (D) Differential expression of ST2, GATA3, T-bet, RORγT, Helios, FoxP3, and IL-18R1 on CD4+FoxP3+ Tregs (gMFI/mean gMFI in both groups) from the pancreas at day 7. (n = 7–8/group.) Two-way ANOVA. *P < 0.05; **P < 0.01. (E) Representative flow cytometry of Helios and ST2 expression on pancreas-isolated CD4+FoxP3+ T cells at day 7 after transfer and frequency of Helios+ Tregs in the pancreas, panLN, and spleen at day 7 after transfer. Two-way ANOVA. ***P < 0.001. (F) Frequency of ST2+ among FoxP3+ Tregs in the pancreas, panLN, and spleen at day 7 after transfer. Two-way ANOVA. ***P < 0.001. (G) Frequency of IL-18R+ among FoxP3+ Tregs in the pancreas, panLN, and spleen at day 7 after transfer. Two-way ANOVA. *P < 0.05.

Since there is evidence that IL-33 drives accumulation of highly suppressive GATA3+ Tregs in the pancreas (33), we asked whether SAR’336 promoted the expansion of GATA3+ Tregs by directly inducing ST2 on Tregs. Using an in vitro approach, we observed that SAR’336 promoted the generation of ST2+ Tregs (Supplemental Figure 6A). Interestingly, SAR’336 also promoted the accumulation of GATA3+ Tregs in the presence of distinct T cell–polarizing conditions (Supplemental Figure 6, B and C), namely Th1 (IL-12 + IL-18), Th2 (IL-4 + IL-33), or Th17 (IL-6 + TGF-β + IL-1β) (54), suggesting that SAR’336 can influence the differentiation of these cells in various inflammatory environments. Indeed, SAR’336 suppressed T-bet expression in Tregs exposed to IL-12 (Supplemental Figure 6, D and E), but not in Teffs (Supplemental Figure 6F), and inhibited RORγT expression in Tregs exposed to IL-6, TGF-β, and IL-1β (Supplemental Figure 6, G and H), verifying a selective effect of SAR’336 on Tregs. Collectively, these observations reveal that, in addition to promoting GATA3 expression, SAR’336 prevents the generation of T-bet+ and RORγT+ Tregs.

Finally, we asked whether SAR’336 influenced the tissue differentiation of early infiltrating Tregs or whether this process happened in the late stages of inflammation. At day 7 after the initial dose, we observed that SAR’336 increased the expression (MFI) of ST2 and Helios and reduced the expression of the Th1-associated IL-18R1, the receptor for IL-18, in pancreatic FoxP3+ Tregs (Figure 5D). We did not observe an increase in GATA3 expression in Tregs by day 7 (Figure 5D), suggesting that SAR’336 initially promotes the expression of ST2, which, in turn, favors the accumulation of GATA3+ Tregs by day 21. While the frequency of Helios+ Tregs was not increased in the pancreas (Figure 5E), we observed a significant increase in the frequency of ST2+Helios+ Tregs in the pancreas and spleen by day 7 after administration of SAR’336 (Figure 5F) and a reduction in the frequency of IL-18R+ Tregs (Figure 5G), highlighting the capacity of SAR’336 to promote the specific generation of protective (33) ST2+ Tregs in the early phase of insulitis in NOD mice.