DF6215, an α-optimized IL-2-Fc fusion, expands immune effectors and drives robust preclinical anti-tumor activity

Summary

DF6215 is a rationally engineered interleukin-2 (IL-2) Fc-fusion protein developed to overcome efficacy and safety limitations of traditional IL-2 cancer immunotherapy. Unlike non-alpha (non-α) IL-2 variants that eliminate CD25 binding and underperform clinically, DF6215 retains moderate IL-2 receptor α (IL-2Rα) affinity while enhancing IL-2Rβγ signaling and extending the half-life via an engineered immunoglobulin (Ig)G1 Fc domain. This design preferentially expands cytotoxic CD8⁺ T cells and natural killer cells over regulatory T cells, resulting in favorable effector-to-regulatory cell ratios, enhanced immune activation, and robust tumor regression in mouse models. In poorly immunogenic tumors, DF6215 synergized with PD-1 blockade to achieve durable responses without added toxicity. Cynomolgus monkey studies confirm DF6215’s pharmacodynamics and favorable safety profile, with no signs of vascular leak syndrome or cytokine release syndrome. These findings position DF6215 as a differentiated IL-2 capable of modulating the tumor microenvironment and achieving potent anti-tumor immunity with improved tolerability, supporting its advancement into clinical trials for solid tumors.

Graphical abstract

Highlights

• •DF6215, an IL-2-based therapy, expands effectors over Tregs via IL-2Rα optimization
• •DF6215 modulates the TME, inducing regressions and outperforming non-α IL-2s
• •DF6215 shows favorable safety and pharmacodynamics in non-human primates
• •DF6215 improves efficacy and reduces toxicity, widening the therapeutic window

Stockmann et al. describe a rationally engineered IL-2-Fc-fusion designed to overcome limitations of traditional IL-2 cancer immunotherapy. DF6215 retains moderate IL-2Rα affinity while enhancing IL-2Rβγ signaling and has an extended half-life. DF6215 drives robust tumor regression in mouse models with a favorable safety profile, both as a monotherapy and in combination with anti-PD-1.

Introduction

Interleukin-2 (IL-2) is a pleiotropic cytokine central to the activation and regulation of the immune system. IL-2 signals through a heterotrimeric receptor complex composed of interleukin 2 receptor alpha (IL-2Rα; CD25), IL-2Rβ (CD122), and the common γ-chain (CD132).^1,2,3,4 The α subunit confers high-affinity binding but lacks signaling capability, while the β and γ chains transduce downstream signals via JAK/STAT pathways.⁵ Regulatory T cells (Tregs) constitutively express the high-affinity IL-2Rαβγ complex, allowing them to efficiently capture IL-2 and suppress immune responses.^6,7,8 In contrast, resting natural killer (NK) cells and memory CD8⁺ T cells express the intermediate-affinity IL-2Rβγ dimer, which supports effector activation independent of IL-2Rα.⁷ However, activated, antigen-experienced CD8⁺ T cells upregulate IL-2Rα, forming the high-affinity IL-2Rαβγ complex.⁷ This upregulation enhances their responsiveness to IL-2 during immune priming and recall responses, contributing to robust clonal expansion and functional persistence. In the context of anti-tumor immunity, this feature is particularly important, as it enables selective amplification of tumor-specific effector T cells that have encountered antigen, thereby strengthening immune surveillance and cytolytic function within the tumor microenvironment (TME).⁹ Thus, IL-2R subunit composition determines cell-type selectivity and influences the balance between immune activation and regulation in therapeutic settings.

IL-2 plays a critical role in the expansion, differentiation, and cytotoxic function of effector immune cells such as CD8⁺ T cells and NK cells, driving potent anti-tumor responses.^2,10,11,12 IL-2 contributes to immune activity by enhancing the cytotoxicity of CD8⁺ T cells and NK cells by promoting granzymes and perforin expression,¹¹ facilitating the differentiation of CD8⁺ T cells upon antigen recognition, leading to memory formation,¹ and reversing CD8⁺ T cell exhaustion, in part through sustained STAT5 signaling and epigenetic remodeling.^13,14,15,16 IL-2 also modulates the TME by inducing the release of interferon gamma (IFN-γ) and upregulating major histocompatibility complex (MHC) expression on tumor and stromal cells, thereby enhancing antigen presentation and recognition of tumor cells.^17,18,19,20 However, the immune-activating effects of IL-2 are intrinsically counterbalanced by its stimulation of Tregs, which express the high-affinity trimeric IL-2R and act to curtail excessive effector cell responses. Consequently, IL-2’s dual roles in promoting immune activation and enforcing immune tolerance underscore the therapeutic complexity of maximizing its anti-tumor potential.^2,6,7

Aldesleukin, a recombinant human IL-2 (rhIL-2) therapy, was approved in the US for the treatment of metastatic renal cell carcinoma (RCC) and metastatic melanoma in 1992 and 1998, respectively.^18,21,22 At high doses, aldesleukin activates and expands cytotoxic lymphocytes, including CD8⁺ T cells and NK cells, promoting effector cell proliferation and function, immune cell infiltration, and antigen presentation, thereby overcoming multiple layers of tumor-associated immunosuppression.²³ However, at low doses, IL-2 preferentially expands Tregs due to their high expression of the IL-2Rα, which allows them to outcompete effector cells for IL-2 and suppress anti-tumor immunity, ultimately reducing therapeutic efficacy.²⁴ Therefore, to fully harness IL-2’s therapeutic potential, it must be administered at high concentrations to drive sustained expansion and activation of cytotoxic effector cells. Achieving this therapeutic threshold is challenging due to IL-2’s inherently short serum half-life, which necessitates frequent and intensive dosing regimens.^25,26 This in turn amplifies the risk of dose-limiting toxicities, particularly vascular leak syndrome (VLS), characterized by plasma extravasation, hypotension, and reduced organ perfusion.²⁷ The underlying mechanism of VLS is unknown, but there is some evidence that IL-2 interacts with endothelial cells via IL-2Rα, which may contribute to its development.²⁸ As a result of these pharmacokinetic (PK) and safety limitations, the clinical utility of aldesleukin remains limited to a small subset of patients who can tolerate the aggressive treatment.^29,30

In an effort to improve the therapeutic window, several “non-alpha” (non-α) IL-2 variants have been developed to avoid IL-2Rα binding to minimize Treg stimulation and reduce toxicity associated with high-dose IL-2 therapy.^31,32 However, bempegaldesleukin (NKTR-214), the first molecule of this class to advance into late-stage clinical development, failed to demonstrate clinical benefit in a pivotal phase 3 trial in combination with nivolumab.^33,34 Emerging evidence indicates that IL-2Rα engagement is essential for maximizing IL-2 synergy with immune checkpoint blockade. Preventing IL-2Rα binding, either via antibody blocking or mutagenesis, eliminates the efficacy of IL-2 in combination with anti-PD-1 therapy.¹⁴ These findings challenge the prior assumption that avoiding IL-2Rα would enhance the therapeutic index. Thus, while non-α IL-2 molecules may avoid VLS and lower Treg stimulation, these benefits come at the cost of a significant loss in anti-tumor activity, particularly in combination with immune checkpoint blockade.⁹

DF6215 is a rationally engineered IL-2 fusion protein developed to overcome the PK, safety, and functional limitations of prior IL-2 therapies. It incorporates amino acid substitutions that reduce, but do not eliminate, IL-2Rα binding while enhancing affinity for IL-2β. These modifications reduce the engagement of Tregs and simultaneously preserve robust, effective engagement of activated and antigen-experienced CD8⁺ T cells, a critical interaction necessary for optimal anti-tumor activity. The IL-2 moiety is fused to a human immunoglobulin (Ig)G1 constant fragment (Fc) engineered to abolish binding to Fcγ receptors while retaining neonatal Fc receptor (FcRn) binding to extend the half-life and systemic exposure while eliminating off-target immune activation. Here, we present the preclinical characterization of DF6215, an IL-2 Fc-fusion protein optimized to retain the therapeutic benefits of IL-2 while minimizing its associated toxicities. These studies highlight DF6215 as an α-optimized IL-2 agonist that can safely enhance anti-tumor immune responses.

Results

DF6215 is a monovalent IL-2 mutein-Fc fusion protein engineered for optimal IL-2R engagement and prolonged half-life

Given that the dual role of IL-2Rα binding in enhancing anti-tumor immune responses and driving Treg expansion and systemic toxicity, we aimed to rationally design an IL-2 variant with a selectively modulated receptor-binding profile to optimize therapeutic benefit while minimizing adverse effects. DF6215 was therefore designed as an IL-2 mutein with attenuated affinity for IL-2Rα to reduce Treg activation and improve safety while preserving signaling on effector populations that express the trimeric IL-2R complex (IL-2Rαβγ), thereby maintaining signaling in key anti-tumor effector cells. In addition, to enhance PK properties and the therapeutic index, DF6215 was engineered as an Fc fusion, known to improve the therapeutic potential of cytokines.³⁵ Importantly, DF6215 was constructed as an asymmetric monovalent IL-2 mutein-Fc fusion to prevent avidity effects that could otherwise enhance IL-2Rα binding, thereby preserving the desired pharmacodynamic (PD) selectivity.

DF6215 is a heterodimeric Fc fusion protein composed of two polypeptide chains, chain A and chain B (Figure 1A). Both chains consist of a truncated IgG1 heavy chain containing only the CH2 and CH3 constant domains. Chain B additionally features a monovalent IL-2 mutein, which is fused to the C terminus of the truncated IgG1 heavy chain via a (G₄S)₃ linker. The IL-2 mutein of DF6215 includes seven targeted amino acid substitutions relative to wild-type (WT) IL-2 (T3A, G27M, I28L, K32D, L72Q, R81D, and C125S). These substitutions were selected to structurally stabilize regions of IL-2’s tertiary structure rather than to directly alter receptor binding interfaces, consistent with previously reported efforts aimed at modulating cytokine activity.^36,37 As a result, DF6215 exhibits reduced affinity for IL-2Rα and enhanced binding to IL-2Rβ. Surface plasmon resonance (SPR) analysis demonstrated that DF6215 exhibits reduced binding affinity to rIL-2Rα, resulting in an approximately 23-fold reduction in binding to the high-affinity trimeric IL-2Rαβγ receptor complex compared to WT IL-2 while increasing binding to rIL-2Rβ by 3.5-fold (Figure 1B).

Figure 1.

DF6215 design and IL-2 receptor binding properties

(A) Cartoon of DF6215’s structural design.

(B) DF6215’s binding to different IL-2R subunits/forms compared to aldesleukin and to human FcRn at pH 6.0 compared to standard IgG1 trastuzumab. K_D, equilibrium dissociation constant. Data are represented as the mean ± SD.

See also Figure S1.

The two chains heterodimerize via a complementary set of seven amino acid substitutions within their CH3 domains, including an engineered interchain disulfide (S-S) bridge that confers structural stability of the heterodimer and prevents homodimer formation.³⁸ Each heavy chain additionally incorporates three amino acid substitutions that abolish Fcγ receptor binding (Fc(si)) while preserving interactions with FcRn,^39,40 allowing for Fc-mediated half-life prolongation compared to hIL-2 without eliciting Fc-effector functions.

DF6215 preferentially activates effector cells over Tregs

The combination of intermediate affinity for IL-2Rα and enhanced binding to IL-2Rβ is a distinguishing feature of DF6215, intended to reduce IL-2Rα-mediated engagement of Tregs and endothelial cells, thereby potentially lowering immunosuppressive signaling and toxicity while preserving engagement with IL-2Rα on activated cytotoxic T cells. To evaluate whether this engineered receptor selectivity translates into functional immune modulation, immune cell activation was measured by quantification of downstream STAT5 signaling of healthy human peripheral blood mononuclear cells (PBMCs) following exposure to DF6215 compared to aldesleukin or non-α IL-2-Fc (Figure 2A). The non-α comparator was engineered in the same format as DF6215, an asymmetric, monovalent IL-2 mutein-Fc fusion with one IL-2 unit per molecule with the same Fc backbone/linker, differing by a single substitution (F42A) that abrogates IL-2Rα binding while preserving IL-2Rβγ signaling relative to WT IL-2.

Figure 2.

DF6215 drives preferential activation of effector lymphocytes in PBMCs and tumor-derived lymphocytes in vitro

(A) Phosphorylation of STAT5 due to DF6215 or comparator IL-2 stimulation in PBMC subsets, as assessed by flow cytometry. CD4⁺_conv T cells are defined as CD4⁺CD25^lowFOXP3⁻. Tregs are defined as CD4⁺CD25⁺FOXP3⁺. The data depicted are from a single donor, representative of data from 9 healthy human PBMC donors.

(B) Phosphorylation of STAT5 due to DF6215 or comparator IL-2 stimulation in activated (CD25⁺) CD8⁺ T cells within TCR-stimulated PBMCs as assessed by flow cytometry focusing. The data depicted are from a single donor, representative of data from 6 healthy human PBMC donors.

(C) Quantified pSTAT5 EC50 (shown as mean ± SEM) in distinct subsets of immune cells shown in (A) and (B) upon stimulation with aldesleukin, DF6215, or non-⍺ IL-2-Fc.

(D) Primary dissociated tumor cells were treated with 0.39 nM of aldesleukin, DF6215, or non-⍺ IL-2 Fc for 20 min prior to assessment of STAT5 phosphorylation. Data points indicate the mean of duplicate wells ± SD from a melanoma biopsy and a lung tumor biopsy. Significance for DF6215 relative to aldesleukin or non-⍺ IL-2 Fc by one-way ANOVA is noted as ∗∗p < 0.01 and ∗∗∗∗p < 0.0001.

DF6215 more effectively activated IL-2Rβγ-expressing naive CD8⁺ T cells and NK cells than either comparator molecule. In contrast, DF6215 demonstrated reduced activation of Tregs compared to that of aldesleukin, but the extent of reduction was less than that seen with the non-α IL-2-Fc, indicating intermediate engagement of CD25. This shift in activation potency toward cytotoxic effector cells and away from immunosuppressive Tregs differentiated DF6215 from both aldesleukin and non-α IL-2-Fc in naive immune cell contexts. In addition, the ability of DF6215 to stimulate T cell receptor (TCR)-activated CD8⁺ T cells, which express high levels of the trimeric high-affinity IL-2Rαβγ, was examined. These activated cells represent tumor antigen-experienced populations that are licensed to mediate anti-tumor immune responses. In anti-CD3/anti-CD28-stimulated human CD8⁺ T cells, DF6215 activated STAT5 signaling comparably to aldesleukin and to a greater degree than non-α IL-2-Fc (Figures 2B and 2C).

To further assess the potential for immune cell activation within a more physiologically relevant tumor setting, the ability of DF6215 to stimulate tumor-infiltrating lymphocytes (TILs) ex vivo was evaluated. Tumor tissue specimens from a patient with melanoma and a patient with non-small cell lung cancer were enzymatically dissociated to generate single-cell suspensions, in which CD8⁺ T cell activation in response to DF6215 was measured (Figure 2D). Consistent with findings in peripheral PBMCs from healthy donors, DF6215 elicited robust STAT5 signaling in tumor-infiltrating CD8⁺ T cells and significantly outperformed aldesleukin and the non-α IL-2-Fc (p < 0.0001). Additionally, DF6215 exhibited a significant increase in potency on activated CD25⁺CD8⁺ T cells (p < 0.01) when compared to the non-α IL-2-Fc.

These data demonstrate that DF6215 preserves the potency of aldesleukin in activating antigen-experienced CD8⁺ T cells that express the high-affinity IL-2R, yet DF6215 induces less Treg activation. This preferential activation profile supports DF6215’s potential to achieve robust anti-tumor immunity by maintaining α-dependent, CD8⁺ T cell-mediated anti-tumor activity, with reduced Treg-mediated immunosuppression.

DF6215 preferentially expands effector cells over regulatory cells in vivo, modulating the TME to drive monotherapy efficacy

Building on in vitro findings that DF6215 preferentially activated CD8⁺ T cells and NK cells over Tregs compared to aldesleukin, we next sought to characterize the immune response in vivo. To this end, naive mice were administered a single low or high dose of DF6215, and peripheral immune cell subsets were quantified over time (Figure 3A). DF6215 led to broad lymphocyte expansion, consistent with systemic IL-2R engagement. Both dose levels led to the proliferation (Ki67⁺) of multiple immune populations, including NK cells, naive CD8⁺ T cells, CD25⁺ CD8⁺ T cells, and Tregs. At the higher dose, CD8⁺ T cells, especially CD25⁺ CD8⁺ T cells, continued to expand beyond levels observed at the lower dose, whereas Treg and NK cell expansion plateaued, showing no further dose-dependent increase. This divergence suggests that exposure level can further enhance the preferential proliferation of activated CD8⁺ T cells relative to Tregs. These data highlight a skewing toward the expansion of cytotoxic effector cells, supporting the potential of DF6215 to preferentially induce anti-tumor immunity.

Figure 3.

DF6215 mediates robust lymphocyte expansion and modulates the tumor microenvironment, resulting in potent anti-tumor activity

(A) Kinetics of Ki-67⁺ immune subset counts in the blood after human (h)IgG1 isotype control or DF6215 administration (0.23 mg/kg in light blue and 1.35 mg/kg in dark blue) in naive BALB/c mice (n = 3/group). Data represent the mean ± SEM.

(B) Efficacy of DF6215 in the mouse CT26 tumor model. Individual tumor volumes, per caliper measurements, are shown with treatment days and frequencies as indicated (vertical dotted lines, QW schedule). The number of complete responders (CRs) in each DF6215 treatment group is noted within each image. n = 10/group. Kaplan-Meier survival curves are shown, with the median survival indicated in brackets (log rank Mantel-Cox test: ∗∗∗p < 0.001 and ∗∗∗∗p < 0.0001). Top: mice were enrolled into treatment groups on day 11, when tumors averaged 218 mm³ in size, and once weekly treatment began. Bottom: mice were enrolled into treatment groups on day 12, when tumors averaged 181 mm³ in size, and once weekly treatment began. See also Figure S2.

(C and D) BALB/c mice (n = 10/group) were engrafted with CT26 and treated i.p. with a single dose of DF6215 (0.675 mg/kg) or hIgG1 isotype after mice were enrolled into treatment groups when tumors averaged ∼218 mm³ in size.

(C) (Left) Absolute cell count quantification of indicated immune subsets in the tumor microenvironment 7 days post-treatment. Data shown are the mean ± SEM. (Right) Immune subset counts in tumors 96 h post-treatment. Significance for DF6215 relative to isotype by unpaired t test was noted as ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001, and ns, not significant.

(D) Cytokines in blood as assessed by a Luminex proinflammatory panel 24 h post-treatment. Significance for DF6215 relative to isotype by one-way ANOVA with Tukey’s test is noted as ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001, and ns, not significant.

To investigate the PK properties of DF6215, we first assessed its circulating half-life. DF6215 exhibited a prolonged half-life relative to aldesleukin (Figure S1), attributable to FcRn-mediated recycling conferred by the Fc domain (Figure 1B), as well as its larger molecular size, which limits renal clearance. Specifically, DF6215 demonstrated a half-life of 11.6–13.5 h, representing a 7- to 8.3-fold improvement compared to the 1.63 h half-life of aldesleukin. This extended exposure profile was designed to support sustained IL-2R engagement while reducing dosing frequency.

We next investigated the in vivo anti-tumor efficacy of DF6215 using the CT26 colon carcinoma model, a syngeneic BALB/c-derived tumor model with relatively high baseline immune cell infiltration, which is sensitive to checkpoint blockade.⁴¹ Mice bearing established subcutaneous CT26 tumors were dosed intraperitoneally (i.p.) weekly (QW) for 4 weeks with escalating doses of DF6215 (0.225, 0.675, 1.35, or 2.7 mg/kg) or with a human IgG1 isotype control. DF6215 induced dose-dependent tumor regressions, achieving complete responses at all doses, ranging from 2 of 10 mice at the low dose to 10 of 10 mice at the highest dose (Figure 3B). DF6215 also resulted in significantly and substantially improved survival for all doses (p < 0.001). The median survival for the low dose was 39 days compared to 19 days for the isotype control, while the majority of mice treated with higher doses survived through the duration of the study, with the median survival not reached by study end. No overt toxicity was observed; however, at higher doses, a transient first-dose body weight decrease was noted, followed by complete recovery, while PD activity remained sustained after repeat dosing (Figure S2), consistent with a class-typical first-dose cytokine effect. These findings indicate that DF6215 achieves potent anti-tumor activity while maintaining a favorable tolerability profile, consistent with an expanded therapeutic window enabled by optimized IL-2Rα binding.

To elucidate the immune mechanisms underlying this efficacy, intratumoral immune profiling was conducted following DF6215 administration in CT26 tumor-bearing mice (Figure 3C). Notably, DF6215 induced significant expansion of CD8⁺ T cells and NK cells compared to the isotype control within the TME, whereas expansion of CD4⁺ T cells, including Tregs, was not significantly different from the isotype control. In addition, DF6215 induced robust expansion of activated IL-2Rα-expressing CD8⁺ T cells (CD25⁺CD8⁺) in the tumor. Consistent with the in vitro data in human cells (Figure 2A), these in vivo findings demonstrate that engineering of the IL-2 moiety in DF6215 has resulted in a variant that preferentially and more potently stimulates activated CD8⁺ effector T cells over Tregs.

To further characterize the systemic effects of DF6215 treatment, we analyzed the peripheral cytokine profile 24 h after administration (Figure 3D). DF6215 administration induced IL-2-associated cytokines, such as IFN-γ and tumor necrosis factor alpha (TNF-α), but at markedly lower levels than those induced by anti-CD3 treatment, which triggers TCR signaling, mimicking the activity of T cell engagers and associated safety, such as cytokine release syndrome. The observed increases in IFN-γ and TNF-α, combined with the absence of detectable IL-10, a cytokine associated with direct Treg engagement or activation of regulatory circuits, further support bias toward CD8⁺ T cells and NK activation. Moreover, the absolute cytokine levels do not indicate excessive immune activation. Importantly, DF6215 did not significantly elevate IL-6 or IL-1β, cytokines commonly implicated in cytokine release syndrome, further supporting a favorable and tolerable safety profile for DF6215.

To contextualize these findings and benchmark performance, we compared DF6215 to a non-α IL-2-Fc fusion, rather than aldesleukin, due to the difficulty of achieving pharmacologically relevant exposures of aldesleukin without toxicity. The design goal for DF6215 was to create an IL-2-based molecule that would be better tolerated than aldesleukin while achieving superior efficacy relative to contemporary non-α IL-2 strategies at equivalent exposures. In addition, by comparing it to a non-α IL-2-Fc, we can directly assess whether the differences in immune cell engagement and activation observed in vitro (Figure 2) translate into differences in in vivo activity. Therefore, the comparison to a non-α IL-2-Fc fusion was chosen as a relevant and appropriate comparator.

Mice bearing established subcutaneous CT26 tumors (∼218 mm³) were dosed i.p. QW for 4 weeks with DF6215 or non-α IL-2-Fc at doses of 0.225 or 0.675 mg/kg or with a human IgG1 isotype control (Figure 4A). DF6215 consistently demonstrated greater anti-tumor efficacy and enhanced survival compared to non-α IL-2-Fc. DF6215 achieved more complete responses and demonstrated significantly better survival compared to non-α IL-2-Fc at each dose level. Collectively, the results indicate that DF6215 achieves superior efficacy relative to non-α IL-2-Fc while maintaining tolerability (Figure S3), consistent with an expanded therapeutic window due to its optimized IL-2Rα binding.

Figure 4.

DF6215 treatment augments anti-tumor activity compared to a non-⍺ IL-2-Fc

(A) Efficacy of DF6215 monotherapy vs. non-⍺ IL-2 Fc monotherapy administered i.p. in the mouse CT26 tumor model. Individual tumor volumes are shown with treatment days and frequencies as indicated (vertical dotted lines). The number of complete responders (CRs) in each treatment group is noted in each image. N = 10/group. Kaplan-Meier survival curves are shown, with the median survival indicated in brackets (∗∗∗p < 0.001, ∗∗p < 0.01, and ∗p < 0.05; log rank Mantel-Cox test). Some of the DF6215 monotherapy data presented in Figure 3 to illustrate the dose response are included here to allow direct comparison with the non-α-binding IL-2-Fc, which was included as a comparator in the same experiment. See also Figure S3.

(B–D) BALB/c mice (n = 10/group) were engrafted with CT26 and treated i.p. with a single dose of hIgG1 isotype, DF6215 (0.675 mg/kg), or non-⍺ IL-2-Fc (0.675 mg/kg).

(B) The effector cell ratios of cells in the tumor microenvironment 96 h post-treatment (∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, and ∗∗∗∗p < 0.0001; one-way ANOVA with Tukey’s test).

(C) Frequencies of CD69⁺ NK and CD8⁺ T cells in the tumor microenvironment 48 h post-treatment. (∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, and ∗∗∗∗p < 0.0001; one-way ANOVA with Tukey’s test).

(D) Frequencies of granzyme B⁺ NK cells and CD8⁺ T cells in the tumor microenvironment 48 h post-treatment (∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, and ∗∗∗∗p < 0.0001; one-way ANOVA with Tukey’s test).

To assess the immunomodulatory effects of DF6215 in CT26 tumor-bearing mice, the ratio of intratumoral NK cells, CD8⁺ T cells, and CD25⁺ CD8⁺ T cells to Tregs was evaluated (Figure 4B). DF6215 treatment resulted in a higher effector-to-Treg ratio compared to both the non-α IL-2-Fc and the isotype control, indicating a preferential expansion of effector populations over immunosuppressive Tregs. Furthermore, while both DF6215 and non-α IL-2-Fc activated (CD69⁺) intratumoral immune cells (Figure 4C), DF6215 enhanced the cytotoxic potential of NK and CD8⁺ T cells above that of non-α IL-2-Fc, as evidenced by elevated granzyme B expression in these populations (Figure 4D).

Taken together, these findings demonstrate that DF6215 not only promotes a favorable intratumoral effector-to-regulatory cell ratio but also potently activates key cytotoxic immune effectors within the TME, highlighting its potential as a promising and differentiated IL-2-based immunotherapy.

DF6215 synergizes with PD-1 blockade to overcome checkpoint resistance

The combinatorial therapeutic potential of DF6215 was further evaluated in the B16F10 melanoma model, a poorly immunogenic (“cold”) tumor known to be resistant to immune checkpoint inhibitors. Given the demonstrated monotherapy efficacy of DF6215 and its ability to modulate the TME, we assessed whether combining DF6215 with anti-PD-1 could enhance anti-tumor responses in this otherwise checkpoint-resistant setting. Another key advantage of an α-active IL-2 molecule is the ability to limit effector T cell exhaustion; therefore, an additional rationale for testing DF6215 in this model was to evaluate whether its engineered IL-2 moiety retains the capacity to synergize with PD-1 blockade.

C57BL/6 mice bearing subcutaneous B16F10 tumors were treated i.p. for 4 weeks with DF6215 or a comparator non-α IL-2-Fc molecule (0.675 mg/kg, QW), anti-PD-1 antibody (200 μg, BIW), or the combination of each IL-2 agent with PD-1 blockade (Figure 5A). Both DF6215 and non-α IL-2-Fc monotherapies delayed tumor growth and extended the median survival compared to isotype controls (p < 0.01), though their effect on tumor growth was limited, consistent with expectations for IL-2 monotherapy in this poorly immunogenic model (Figure 5A). However, when combined with PD-1 blockade, both IL-2 agents showed enhanced anti-tumor activity; however, the combination of DF6215 and PD-1 blockade achieved superior outcomes compared to the combination of non-α IL-2-Fc and PD-1. Specifically, the DF6215/PD-1 combination yielded a 7/9 complete response rate, compared to 2/9 for the non-α IL-2-Fc/PD-1 combination, and median survival improvements vs. DF6215 alone (p < 0.01), PD-1 alone (p < 0.001), and isotype controls (p < 0.01), demonstrating clear therapeutic synergy and an advantage with DF6215’s α-active design.

Figure 5.

DF6215 combines well with checkpoint blockade

(A) C57BL/6 (n = 9/group) were inoculated with B16F10 melanoma cells and enrolled into treatment groups on day 10, when tumors averaged ∼85 mm³ in size. Mice were dosed i.p. with isotype controls (hIgG1 and rIgG2a), DF6215, or non-⍺ IL-2 Fc at 0.675 mg/kg once weekly, anti-PD-1 at 200 μg twice weekly, or the two in combination. Individual tumor volumes are shown with treatment days and frequencies as indicated (vertical dotted lines, QW). The number of complete responders (CRs) in each treatment group is noted in each image.

(B) Kaplan-Meier survival curves are shown, with median survival indicated in brackets (∗∗∗∗p < 0.000, ∗∗∗p < 0.001, ∗∗p < 0.01, and ∗p < 0.05; log rank Mantel-Cox test). Significance for survival curves: ∗p < 0.05 for DF6215 vs. anti-PD-1 monotherapy; ∗∗p < 0.01 for DF6215 vs. monoclonal antibody (mAb) isotypes, non-⍺ IL-2 Fc vs. mAb isotypes, DF6215 combination vs. mAb isotypes, DF6215 combination vs. DF6215 monotherapy, and DF6215 combination vs. non-⍺ IL-2 Fc combination; ∗∗∗p < 0.001 for DF6215 combination vs. anti-PD-1 monotherapy and non-⍺ IL-2 Fc combination vs. mAb isotypes; and ∗∗∗∗p < 0.0001 for non-⍺ IL-2 Fc combination vs. mAb isotypes.

See also Figure S4.

Importantly, combination treatment remained well tolerated, with no additional adverse effects noted in the combination groups compared to DF6215 monotherapy (Figure S4). Of note, approximately 40% of mice with complete responses to the DF6215/PD-1 combination developed vitiligo, a phenotype consistent with robust and systemic CD8⁺ T cell-driven immunity. These mice remained healthy and alert and steadily gained weight throughout the study.

Collectively, these findings underscore the ability of DF6215 to synergize with PD-1 blockade in cold tumor models and underscore its superior combinatorial efficacy relative to a non-α IL-2-Fc comparator. This synergy drives durable tumor rejection even in checkpoint-refractory tumors without exacerbating toxicity.

DF6215 exhibits a reduced potential for systemic cytokine release in comparison to aldesleukin in human in vitro assays

Cytokine release induced by IL-2-based therapeutics is a key contributor to treatment-related toxicities; therefore, understanding the cytokine profile of DF6215 is essential for characterizing the potential for cytokine-related toxicities. We evaluated the in vitro cytokine release profile of DF6215 from human PBMCs using both soluble and immobilized (wet-coated) assay formats to assess the potential for cytokine-related toxicities in patients. The objective of this study was to compare the cytokine induction potential of DF6215 to that of rhIL-2 (aldesleukin) at molar-equivalent concentrations.

At concentrations ≥ 3 nM, DF6215 induced the release of cytokines associated with IL-2 pharmacology, including IFN-γ, TNF-α, and IL-5. While most cytokines exhibited comparable levels between DF6215 and aldesleukin, a subset of cytokines, notably IL-6, TNF-α, and IL-1Rβ, displayed reduced release even at higher concentrations of DF6215 compared to aldesleukin (Figure 6). This reduced cytokine release profile for DF6215 was consistently observed across multiple donors and both assay formats and aligned with observations from murine studies (Figure 3D), suggesting that DF6215 may offer an improved safety profile with respect to cytokine-mediated toxicities.

Figure 6.

DF6215 triggers less cytokine release from human PBMCs than aldesleukin in vitro

Freshly isolated human peripheral blood mononuclear cells (PBMCs) from 5 healthy donors were stimulated with indicated soluble molecules at 37°C for 48 h. For positive controls, cells were treated with coated anti-CD3 at 2 μg/well. Secreted cytokines in the supernatant were analyzed using a customized 12-plex Luminex kit (MILLIPLEX): granulocyte colony-stimulating factor (G-CSF), IFN-γ, IL-1RA, IL-1β, IL-2, IL-5, IL-6, IL-10, IL-12p70, MCP-1 (CCL2), MIP-1β (CCL4), and TNF-α. Samples were run in triplicate for each donor; data are presented as the means from each donor, with the overall mean values across donors ± SD marked with a dash and error bars, respectively. The significance for the cytokine response was determined by a one-way analysis of variance (ANOVA) multiple comparisons test (Tukey’s). The upper limit of detection is indicated by the dotted line.

DF6215 demonstrates PD activity and a favorable safety profile in cynomolgus monkeys

Following the promising efficacy and mechanistic data in murine tumor models, the PK, PD responses, and the safety profile of DF6215 were evaluated in cynomolgus monkeys to better capture the complexities of the human immune response in a more translationally relevant model.

The in vitro potency of DF6215 was evaluated in cynomolgus PBMCs using intracellular pSTAT5 detection by flow cytometry. DF6215 induced dose-dependent activation of IL-2R-expressing immune cells (Figure S5A). Compared to aldesleukin, DF6215 maintained strong activation of IL-2Rβγ-expressing effector cells, including naive CD8⁺ T cells and NK cells, while demonstrating reduced activation of Tregs and CD4⁺ T cells. This selective immune activation profile mirrors the pattern observed in human PBMCs, supporting the relevance of the cynomolgus model and reinforcing DF6215’s ability to preferentially stimulate cytotoxic effectors while minimizing immunosuppressive responses.

Male and female cynomolgus monkeys were administered via intravenous bolus either vehicle or DF6215 at 30, 100, or 300 μg/kg once every 2 weeks for a total of 3 administrations. Dose-related increases in DF6215 exposure were observed. PK analysis showed an extended serum half-life of 4.4–7.9 h, consistent with binding of the Fc domain to FcRn (Figures 7A and 1B), substantially longer than the ∼85-min half-life of aldesleukin.^20,23,25,26 Treatment-emergent anti-DF6215 antibodies (ADAs) were detected at doses of ≥100 μg/kg but did not impact exposure or peripheral PD, and no ADA-related toxicities were observed. DF6215 was well tolerated up to 300 μg/kg. No treatment-related effects were observed on neurological function, electrocardiograms, respiration, or behavior. Histopathology showed no macroscopic findings, and DF6215-related microscopic changes were limited to mononuclear or mixed-cell infiltrates in various tissues at doses of ≥30 μg/kg. Importantly, there was no evidence of VLS in monkeys, as demonstrated by the lack of blood pressure changes after repeat dosing and the absence of microscopic effects, such as tissue edema and fluid buildup in cavities.

Figure 7.

Improved safety and PD response of DF6215 in cynomolgus monkeys

(A) Pharmacokinetics of DF6215 after a single bolus in cynomolgus monkeys. n = 3–5/sex/group. Data represent the mean ± SEM.

(B) Counts of lymphocytes and eosinophils over time in blood of cynomolgus monkeys after multiple doses of DF6215 (dosing depicted by vertical dotted lines). n = 2/sex/group. Data represent the mean ± SEM.

(C) Kinetics of Ki-67⁺ effector cells in the blood of cynomolgus monkeys after multiple doses of DF6215 (dosing depicted by vertical dotted lines). n = 2/sex/group. Data represent the mean ± SEM.

See also Figure S5.

Supportive of these GLP findings, a non-GLP telemetry study in cynomolgus monkeys showed dose-dependent increases in body temperature, heart rate (≥30 μg/kg), and QTc intervals (≥100 μg/kg), with no changes in blood pressure or evidence of hypotension. All observed changes were minimal in magnitude, fully reversible, and not considered adverse, further reinforcing DF6215’s favorable cardiovascular and physiologic safety profile.

Lymphocyte kinetics showed a transient decline at 24 h post-dose, followed by expansion that peaked around day 8 and returned to baseline by day 14 after each dose (Figure 7B). This expansion was accompanied by a dose-dependent increase in proliferation (Ki67⁺) across multiple immune subsets, including CD4⁺ cells, CD8⁺ cells, Tregs, and NK cells (Figure 7C). These patterns closely mirrored those observed in murine models, reinforcing the translational relevance of DF6215-induced immune activation. Cytokine profiling revealed transient increases in IL-1Rβ, IL-5, and MCP-1 without inducing CRS-associated cytokines such as IL-6 or TNF-α (Figure S5B). Peripheral eosinophil expansion was observed with delayed kinetics relative to lymphocyte expansion, peaking at approximately day 10 post-first dose (Figure 7B), consistent with known IL-2 pharmacology.⁴² In contrast to findings described for WT IL-2 in cynomolgus monkeys,⁴³ no histological evidence of eosinophilic infiltration, VLS, or hypersensitivity was observed.

DF6215-related findings were considered related to IL-2 activity and consistent with the expected pharmacology of DF6215.

Discussion

This study presents DF6215 as a next-generation IL-2 Fc fusion protein with a mechanistic and therapeutic profile that distinguishes it from both historical and contemporary IL-2-based therapies. DF6215 was rationally engineered to maintain moderate IL-2Rα binding, critical for activating antigen-experienced CD8⁺ T cells, while enhancing IL-2Rβγ stimulation to drive robust lymphocyte activation and limit Treg stimulation. In multiple murine tumor models, DF6215 demonstrated monotherapy responses and significantly improved the efficacy of PD-1 blockade in immunologically cold tumors, with associated remodeling of the TME. In cynomolgus monkey studies, DF6215 demonstrated an extended half-life, dose-dependent lymphocyte expansion, and a favorable safety profile, including no evidence of vasculature leak syndrome and no observations of CRS-associated cytokine release, hypotension, evidence of tissue edema, or any adverse toxicity in the lung or other target organs, providing a strong foundation for clinical translation.

DF6215 is a promising alternative to traditional IL-2 therapies, engineered to overcome the limitations of native IL-2, which include poor PK, toxicity, and preferential expansion of Tregs via high-affinity IL-2Rα engagement. A major historical limitation of IL-2 therapy has been dose-limiting toxicities such as VLS, which have restricted its clinical use. DF6215 has been designed to mitigate these risks by modulating IL-2Rα affinity to avoid endothelial activation. In preclinical studies, DF6215 did not induce hypotension or signs of vascular leak, supporting its favorable safety profile.

DF6215 preserves IL-2Rα engagement, therefore retaining potency on antigen-experienced CD8⁺ T cells, an effector subset central to durable anti-tumor immunity. DF6215 preferentially expands cell populations with high IL-2R expression, such as NK cells and activated CD8⁺ T cells, whereas conventional CD4⁺ T cells, which express lower levels of all IL-2R subunits, receive weaker IL-2R signaling and thus exhibit less expansion following DF6215 treatment. While both antigen-experienced CD8⁺ T cells and Tregs can express high levels of IL-2Rα (CD25), intrinsic differences in IL-2 signal transduction likely underlie their differential responses: Tregs exhibit elevated PTEN, which dampens PI3K/AKT activation downstream of IL-2R, limiting proliferative and metabolic responses compared with CD8⁺ T cells that more fully exploit both JAK/STAT and PI3K/AKT pathways.^44,45 Such signaling differences could bias toward effector CD8⁺ T cell expansion and activation in the TME following DF6215 treatment, despite high-affinity receptor engagement on both cell types. This allows DF6215 to harness the full therapeutic potential of IL-2 while maintaining a favorable safety profile. In contrast, non-α IL-2 molecules, such as bempegaldesleukin (NKTR-214), were engineered to reduce IL-2Rα binding and selectively stimulate IL-2Rβγ-expressing effector cells to minimize Treg expansion.⁴⁶ However, despite promising preclinical rationale, NKTR-214 failed in multiple clinical trials due to insufficient efficacy. Other non-α designs, such as MDNA11 and nemvaleukin alfa (ALKS 4230), seek to modify receptor selectivity using sequence mutagenesis or fusion-based steric occlusion to eliminate engagement of IL-2Rα, respectively.^47,48 While these approaches can improve safety, they often compromise efficacy by reducing the potency or breadth of immune activation. Notably, nemvaleukin alfa was recently discontinued from clinical development after failing to demonstrate sufficient clinical activity in both ovarian cancer and melanoma trials. Compared to non-α IL-2 variants, DF6215 demonstrates superior preclinical efficacy and more preferential effector cell activation, which sets the stage for clinical benefit and differentiation.

Several alternative strategies are in development to overcome the limitations of rIL-2; however, each comes with trade-offs. Conditionally active IL-2 molecules, such as WTX-124,⁴⁹ aim to restrict IL-2 activity to the TME via protease-dependent unmasking. However, suboptimal cleavage efficiency can require high systemic dosing, increasing the risk of non-specific activation and associated toxicities. Similarly, tumor-targeted IL-2 fusion proteins, including CUE-101⁵⁰ and NHS-IL2,⁵¹ employ antibody-based domains to preferentially localize IL-2 to tumors, thereby limiting systemic exposure. These approaches rely heavily on the level and homogeneity of target antigen expression, which can vary within tumors and among patients, and often necessitates attenuation of IL-2 potency to achieve the desired tumor biodistribution and maintain tolerability. Meanwhile, cis-targeted IL-2 molecules direct stimulation to specific effector subsets, such as PD-1⁺ immune cells (IBI363⁵² or eciskafusp alfa⁵³) or CD8⁺ T cells (AB248⁵⁴) to enhance selectivity and minimize off-target effects. However, this restricted targeting can reduce the breadth of anti-tumor lymphocyte activation and, like tumor-targeting strategies, frequently requires the attenuation of IL-2 potency to prevent systemic toxicity. STK-012 is an IL-2Rα/β agonist that retains IL-2Rα binding and is designed to selectively stimulate CD25⁺ T cells, but this would include Tregs in addition to activated CD8⁺ T cells. Another potential limitation is that STK-012 does not meaningfully expand or activate NK cells and thus may not benefit from NK cell-mediated anti-tumor activity or fully synergize with therapeutic monoclonal antibodies that rely on antibody-dependent cellular cytotoxicity.

DF6215 overcomes these limitations by optimizing IL-2Rα binding to maintain selective effector engagement while avoiding excessive Treg stimulation and toxicity. DF6215 is active in the tumor, stimulating a broad range of anti-tumor lymphocytes without relying on protease activity, expression of a tumor-associated antigen, or attenuation of the IL-2 moiety. The ability to combine safety, systemic exposure, and preferential stimulation of anti-tumor lymphocytes without conditional activation measures sets DF6215 apart from other engineered IL-2s.

In summary, DF6215 is a mechanistically differentiated IL-2 agonist that modulates the TME, promotes immune activation, and balances efficacy with safety through selective receptor engagement. These data support the ongoing phase 1/2 clinical evaluation of DF6215 in patients with advanced solid tumors (ClinicalTrials.gov: NCT06108479).

Limitations of the study

This manuscript describes preclinical data showing that DF6215 clearly differentiates from the commercial agent aldesleukin and a representative non-α IL-2 that is fused to Fc. However, which modifications within DF6215 that deviate from WT IL-2 (i.e., reduced binding to IL-2Rα and enhanced binding to IL-2Rβ or Fc fusion for an extended half-life) confer specific properties is less clear. Given the cooperative nature of IL-2’s trimeric receptor, precisely parsing the individual subunit contributions is experimentally challenging. Mechanistically, IL-2Rα is a non-signaling capture subunit, whereas signaling occurs exclusively through IL-2Rβγ. WT IL-2 can engage three IL-2R states: α alone (no signaling), the βγ dimer (intermediate affinity and signals), and the αβγ trimer (highest apparent affinity due to avidity, signals via βγ). Because trimer assembly is cooperative, the biology is not the sum of “α-effects” plus “βγ-effects.” Despite this, our evidence converges from orthogonal readouts (SPR binding, subset-resolved pSTAT5 signaling, ex vivo TIL activity, and matched-exposure in vivo studies) showing that DF6215’s α-attenuated/β-enhanced profile broadens effector recruitment and cytotoxic function while maintaining potency on antigen-experienced IL-2Rα+ CD8 T cells.

Resource availability

Lead contact

Further information and requests for resources and reagents should be directed to and will be fulfilled by the contact, Ann F. Cheung (ann.cheung@dragonflytx.com).

Materials availability

All reagents listed in the key resources table that are not commercially available may be provided upon request. Reagent requests will require internal review and approval, and requesters may be required to complete a materials transfer agreement (MTA), liability waiver, and/or research agreement, as appropriate.

Data and code availability

This paper does not report original code. Data reported in this paper and any additional information required to reanalyze the data reported herein can be made available from the lead contact upon request.

Acknowledgments

This study was supported by Dragonfly Therapeutics. We thank our colleagues and advisors at Dragonfly Therapeutics for their discussions and collaboration, in particular, Jaafar Haidar, Zong Sean Juo, Sara Basinski, and Asya Grinberg for their contributions to this work. We thank Chris Garcia (Stanford University) for his pioneering work on IL-2, which laid the groundwork for this research and informed the development of DF6215.

Author contributions

A.P.S. designed and performed the in vivo studies, performed the in vitro studies, performed the statistical analysis, and had unrestricted access to all data. S.V. conceived, designed, and directed the study; oversaw the statistical analyses; and had unrestricted access to all data. L.H. performed the in vitro and in vivo assays and performed the statistical analyses. C.-S.H. and J.M. engineered and produced DF6215 and non-α IL-2-Fc. D.F. designed and performed the SPR assays and performed the statistical analysis. P.K. designed and directed the pharmacology and toxicology studies in cynomolgus monkeys. E.G. directed the study. D.T. and S.W.H. wrote, reviewed, and edited the manuscript; performed the statistical analysis; and had unrestricted access to all data. A.F.C. and N.W. conceptualized the study, reviewed and edited the manuscript, and had unrestricted access to all data. All authors agreed on the content of the manuscript.

Declaration of interests

All authors were employees of Dragonfly Therapeutics, Inc. S.V. is currently affiliated with Moderna (Cambridge, MA, USA). L.H. is currently affiliated with Eastern Michigan Homer Stryker School of Medicine (Kalamazoo, MI, USA). C.-S.H. is currently affiliated with BioMedicine Design, Pfizer, Inc. (Cambridge, MA, USA). Dragonfly Therapeutics developed the clinical IL-2-Fc candidate DF6215, currently in phase 1/2 (ClinicalTrials.gov: NCT06108479) and owns a patent application, published as International Patent Publication WO/2025/007037, that is related to the subject matter of this manuscript.

STAR★Methods

Key resources table

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
InVivoMAb human IgG1 isotype control	Bio X cell	Cat# BE0297; RRID: AB_2687817
InVivoMAb anti-mouse PD-1 (CD279)	Bio X cell	Cat# BE0146; RRID: AB_10949053
InVivoMAb rat IgG2a isotype control, anti-trinitrophenol	Bio X cell	Cat# BE0089; RRID: AB_1107769
InVivoMAb anti-mouse CD3ε	Bio X cell	Cat# BE0001-1, RRID: AB_1107634)
PE Mouse IgG1, κ Isotype Ctrl (clone MOPC-21)	BioLegend	Cat# 400112; RRID: AB_2847829
PE anti-human CD25 (clone M-A251)	BioLegend	Cat# 356104, RRID: AB_2561861
R-Phycoerythrin-AffiniPure F(ab')2 Fragment Goat Anti-Human IgG, Fcγ Fragment Specific (secondary antibody)	Jackson ImmunoResearch Labs	Cat# 109-116-170, RRID: AB_2337681
Human TruStain FcX™ (Fc Receptor Blocking Solution) (clone 3G8)	BioLegend	Cat# 422302, RRID: AB_2818986
AF488 anti-human CD45RA (clone HI100)	BioLegend	Cat# 304114, RRID: AB_528816
PE/Dazzle™ 594 anti-T-bet (clone 4B10)	BioLegend	Cat# 644828, RRID: AB_2565677
BV421 anti-human CD25 (clone M-A251)	BioLegend	Cat# 356114, RRID: AB_2562164
BV510 anti-human CD4 (clone SK3)	BioLegend	Cat# 344634, RRID: AB_2566017
BV605 anti-human CD3 (clone SK7)	BioLegend	Cat# 344836, RRID: AB_2565825
BV785 anti-human CD8 (clone SK1)	BioLegend	Cat# 344740, RRID: AB_2566202
AF647 mouse anti-Stat5 (clone 47/Stat5 [pY694])	BD Biosciences	Cat# 612599, RRID: AB_399882
PE-Cy5 anti-human FOXP3 (clone 236A/E7), eBioscience	Thermo Fisher Scientific	Cat# 15-4777-42, RRID: AB_2811750
V450 anti-human FOXP3 (clone 259D/C7)	BD Biosciences	Cat# 560459, RRID: AB_1645591
BV605 anti-human CD3 (clone SP34-2)	BD Biosciences	Cat# 562994, RRID: AB_2737938
FITC anti-NHP CD45 (clone D058-1283)	BD Biosciences	Cat# 557803, RRID: AB_396879
BB700 anti-human CD25 (clone M-A251)	BD Biosciences	Cat# 566447, RRID: AB_2744335
PE-Cy7 anti-human CD4 (clone L200)	BD Biosciences	Cat# 560644, RRID: AB_1727474
Purified anti-mouse CD16/32 (mouse Fc block) (clone 93)	BioLegend	Cat# 101302, RRID: AB_312801
BUV395 mouse anti-Ki-67 (clone B56)	BD Biosciences	Cat# 564071, RRID: AB_2738577
BUV496 anti-mouse CD4 (clone GK1.5)	BD Biosciences	Cat# 612952, RRID: AB_2813886
BUV805 anti-mouse CD49b (clone HMα2)	BD Biosciences	Cat# 741962, RRID: AB_2871270
eFlour450 anti-mouse PD-1 (clone J43)	Thermo Fisher Scientific	Cat# 48-9985-82, RRID: AB_2574139
BV510 anti-mouse CD8a (clone 53–6.7)	BioLegend	Cat# 100752, RRID: AB_2563057
BV605 anti-mouse CD19 (clone 6D5)	BioLegend	Cat# 115540, RRID: AB_2563067
BV785 anti-mouse CD3 (clone 17A2)	BioLegend	Cat# 100232, RRID: AB_2562554
FITC anti-human/mouse Granzyme B (clone GB11)	BioLegend	Cat# 515403, RRID: AB_2114575
PerCP anti-mouse CD45 (clone 30-F11)	BioLegend	BioLegend Cat# 103130, RRID: AB_893339
BB700 anti-Mouse CD25 (clone PC61)	BD Biosciences	Cat# 566498, RRID: AB_2744345
PE/Dazzle™ 594 anti-mouse CD314 (NKG2D) (clone CX5)	BioLegend	Cat# 130214, RRID: AB_2728148
PE-Cy5 anti-mouse FOXP3, ebioscience (clone FJK-16s)	Thermo Fisher Scientific	Cat# 15-5773-82, RRID: AB_468806
PE-Cy7 anti-mouse CD69, ebioscience, (clone H1.2F3)	Thermo Fisher Scientific	Cat# 25-0691-82, RRID: AB_469637
APC anti-mouse CD366 (TIM3) (clone RMT3-23)	BioLegend	Cat# 119706, RRID: AB_2561656
AF700 anti-mouse/human CD11b (clone M1/70)	BioLegend	Cat# 101222, RRID: AB_493705
BV421 mouse anti-Ki67 (clone B56)	BD Biosciences	Cat# 562899, RRID: AB_2686897
BV510 anti-human CD69 (clone FN50)	BioLegend	Cat# 310936, RRID: AB_2563834
BV711 anti-human CD4 (clone L200)	BD Biosciences	Cat# 563913, RRID: AB_2738484
BV786 anti-human CD45 (clone D058-1283)	BD Biosciences	Cat# 563861, RRID: AB_2738454
FITC anti-human CD8 (clone SK1)	BioLegend	Cat# 344704, RRID: AB_1877178
PE-Cy7 anti-human CD159a (clone Z199)	Beckman Coulter	Cat# B10246, RRID: AB_2687887
AF647 anti-human FoxP3 (clone 206D)	BioLegend	Cat# 320114, RRID: AB_439754
AF700 anti-human CD20 (clone 2H7)	BD Biosciences	Cat# 560631, RRID: AB_1727447
PE anti-human CD25 (clone M-A251)	Biolegend	Cat# 356104, RRID: AB_2561861
Live/Dead Near-IR	Invitrogen	Cat# L10119
Zombie NIR™ Fixable Viability Kit	BioLegend	Cat# 423106
Ultra-LEAF™ Purified anti-human CD3	BioLegend	Cat# 317325, RRID: AB_11147370
Ca2 Silent Isotype control (anti-human HER2 hIgG1 Fc(si))	This paper	N/A
Chemicals, peptides, and recombinant proteins
Proleukin® (aldesleukin)	Clinigen, Inc	NDC: 76310-022-01
10 mM sodium acetate	Cytiva	Cat# BR100351
10X HBS-EP+	Cytiva	Cat# BR1006-69
PMSF	Millipore Sigma	Cat# 10837091001
DF6215	This paper	Additional sequence details can be found in patent; WO/2025/007037
Non-alpha IL-2 Fc	This paper	Contains the F42A mutation to abrogate CD25 binding
Percoll®	Millipore Sigma	Cat# GE17-0891-01
Ficoll® Paque PLUS	Millipore Sigma	Cat# GE17-1440-02
Gibco™ 10X PBS, pH 7.4	Thermo Fisher Scientific	Cat# 70011-044
BD Pharmingen™ Stain Buffer (FBS)	BD Biosciences	Cat# 554656
Gibco™ ACK Lysis Buffer	Thermo Fisher Scientific	Cat# A1049201
10X RBC Lysis Buffer (multi species), eBioscience	Thermo Fisher Scientific	Cat# 00-4300-54
Fixation/Permeabilization Concentrate (4X), eBioscience	Thermo Fisher Scientific	Cat# 00-5223
Fixation/Permeabilization Diluent, eBioscience	Thermo Fisher Scientific	Cat# 00-8333
Stop Solution for TMB Substrates	Thermo Fisher Scientific	Cat# N600
Critical commercial assays
BD Pharmingen™ Transcription Factor Phospho Buffer Set	BD Biosciences	Cat# 563239
Bio-Plex® Cell Lysis Kit	BioRad	Cat# 171304011
Pierce™ BCA Protein Assay Kit	Thermo Fisher Scientific	Cat# 23225
Invitrogen™ eBioscience™ Foxp3/Transcription Factor Staining Buffer Set	Thermo Fisher Scientific	Cat# 00-5523-00
Tumor Dissociation Kit, mouse	Miltenyi Biotec	Cat# 130-096-730
ELISA MAX™ Deluxe Set Human IL-2	BioLegend	Cat# 431816
Cytokine & Chemokine Convenience 26-Plex Mouse ProcartaPlex™ Panel 1	Thermo Fisher Scientific	Cat# EPXR260-26088-901
MILLIPLEX® Human Cytokine/Chemokine/Growth Factor Panel A – Immunology Multiplex Assay	Millipore Sigma	Cat# HCYTA-60K-12C
Experimental models: Cell lines
B16F10	American Type Culture Collection	CRL-6475; RRID: CVCL_0159
CT26	American Type Culture Collection	CRL-2638; RRID: CVCL_7256
Experimental models: Organisms/strains
Mouse: C57BL/6J	Jackson Laboratory, Sacramento CA	000664; RRID: IMSR_JAX:000664
Mouse: BALB/cJ	Jackson Laboratory, Sacramento CA	000651; RRID: IMSR_JAX:000651
Cynomolgus monkeys	Charles River Laboratories	N/A
Software and algorithms
WinNonlin v8.3	Certara	https://www.certara.com/software/phoenix-winnonlin/
GraphPad Prism v9.0	Graphpad	https://www.graphpad.com
FlowJo	BD	https://www.flowjo.com
Other
Human IL-2 R alpha, His, Avitag	ACRO Biosystems	Cat# ILA-H82E6
Human IL-2 R beta, His, Avitag	ACRO Biosystems	Cat# ILB-H82E3
Human IL-2RB&IL-2RG Heterodimer Protein, His tag & Twin Strep tag	ACRO Biosystems	Cat# ILG-H5283
Human IL-2RB&IL-2RG&IL-2RA Protein, His tag & Twin Strep tag	ACRO Biosystems	Cat# ILG-H52W9
Twin-Strep-tag Capture Kit	IBA Life Sciences	Cat# 2-4370-010
Amine Coupling Kit	Cytiva	Cat# BR100050
Series S Sensor Chip CM5	Cytiva	Cat# 29-1496-03
Streptavidin planar chip	Xantec	Cat# SCBS SAP
Dynabeads™ Human T-Activator CD3/CD28 for T cell Expansion and Activation	Thermo Fisher Scientific	Cat# 11131D
UltraComp Compensation beads	Invitrogen	Cat# 01-2222-42
“The Big Easy” EasySep™ Magnet	Stemcell	Cat# 18001
FACS Celesta Flow Cytometer	BD Biosciences	RRID: SCR_019597
Human whole blood	BioIVT	Cat# HUMANWBK2-0110581
Cytek Aurora Spectral Analyzer	Cytek	RRID: SCR_019826
xMAP Intelliflex DR-SE System	Luminex Corp.	RRID: SCR_023348

Experimental models and study participant details

Mice

Six to nine weeks old female BALB/cJ or C57BL/6J mice were obtained from Jackson Laboratory and housed in individually ventilated cages with food and water provided ad libitum. Male mice were not used in any studies due to increased aggression and subsequent housing requirements. All animal protocols were approved by the Institutional Animal Care and Use Committee (IACUC) and all procedures were in accordance with approved guidelines. A complete response (CR) was defined as no palpable tumor detected.

Cynomolgus monkeys

Studies in male and female Cambodian cynomolgus monkeys were performed at Charles River Laboratories. All animals were 2–3 years of age, weighing between 1.9 kg and 2.7 kg at study initiation. Treatment groups comprised of equal number or male and female animals. Animals were dosed by bolus intravenous injection on study days 1 and 15. The protocol and any amendment(s) or procedures involving the care and use of animals in these studies were reviewed and approved by Charles River Laboratories (USA) IACUC before conduct. Housing set-up was as specified in the USDA Animal Welfare Act (Code of Federal Regulations, Title 9) and as described in the Guide for the Care and Use of Laboratory Animals (NRC, current edition). Animals were group-housed (up to 3 animals of the same sex and same dosing group together). The animals were separated during designated procedures/activities or as required for monitoring and/or health purposes as deemed appropriate by the Study Director and/or Clinical Veterinarian. All animals used on study had documentation to confirm 1 negative tuberculosis (TB) test. Diet consisted of PMI Nutrition International Certified Primate Chow No. 5048 and was supplemented with fruit or vegetables at least 2–3 times weekly.

Cell lines

The mouse melanoma cell line B16F10 and colon carcinoma cell line CT26 were obtained from ATCC. The subline CT26–20.7, which ectopically-expresses the tyrosinase-related protein 1 (Tyrp1) was used for in vivo studies. The CT26-Tyrp1 in vivo model demonstrated a similar tumor growth profile as the parental CT26 cell line (Figure S2) signifying similar immunogenicity; therefore, for simplicity, it is referred to herein as “CT26”. All tumor cell lines were tested for mycoplasma contamination prior to implantation (IDEXX BioResearch). Tumor cell lines were maintained in Dulbecco’s modified Eagle’s medium (B16F10) or Roswell Park Memorial Institute medium (CT26) supplemented with 10% heat-inactivated fetal bovine serum (Thermo Fisher), 1% Glutamax (Thermo Fisher), 1% penicillin-streptomycin (Thermo Fisher).

Method details

IL-2 fusion protein

DF6215, wtIL-2-Fc and non-alpha IL-2-Fc were produced in Chinese Hamster Ovary (CHO) cells. The clarified CHO harvest was captured and concentrated by standard preparative protein A chromatography methods. Further purification using ion-exchange chromatography removed process and product-related impurities to yield the highly purified IL-2 fusion proteins.

SPR binding experiments

IL-2Rα and IL-2β binding experiments

Recombinant IL-2Rα-His-AviTag-biotin and IL-2Rβ-His-AviTag-biotin were diluted to 50 ng/mL in Biacore running buffer (1X HBS-EP+) and injected for 140 s at 5 μL/min across a planar SPR chip surface with amine coupled streptavidin to a capture level of 40 RU. DF6215 and aldesleukin were buffer exchanged into running buffer and titrated from 10000 nM–4.9 nM and 1000 nM–0.49 nM, respectively, in 2-fold dilutions across the IL-2Rα chip surface. DF6215 and aldesleukin were titrated from 5000 nM–2.4 nM in 2-fold dilutions across the IL-2Rβ chip surface. Association was monitored for 30 s, and dissociation was monitored for 120 s, both at 30 μL/min. The experiment was run at 25°C and each interaction fully returned to baseline and thus no regeneration was used between cycles. A series of 4 blank start-up cycles was performed prior to sample analysis to stabilize the chip surface. Two buffer cycles at the start of each concentration series and a streptavidin reference flow cell with no captured IL-2Rα or IL-2Rβ were used for double-referencing during data analysis. Raw data were fitted to a 1:1 binding model with global analysis using Biacore Insight Evaluation Software. All reported values were rounded to 1 decimal place.

IL-2Rαβγ binding experiments

An SPR chip designed to capture Twin-Strep-tagged proteins was prepared by amine coupling Strep-Tactin XT to the carboxymethyl dextran matrix of a CM5 chip surface to a surface density of ∼3000 RU. Twin-Strep tagged IL-2Rαβγ was diluted in Biacore running buffer and injected across the Strep-Tactin XT surface for 60 s at a flow rate of 10 μL/min to a capture density of 193–211 RU. DF6215 and aldesleukin were buffer exchanged into running buffer and were titrated in 3-fold dilutions from 1.85–0.008 nM across IL-2Rαβγ. Association was monitored for 300 s, and dissociation was monitored for 60 s for all concentrations except the top concentration, which was monitored for 7200 s. Association and dissociation phases were both run at 30 μL/min. Each cycle was completed with 3 × 60-s pulses of 3M GuHCl at 30 μL/min. The experiment was run at 25°C. Two buffer cycles at the start of each concentration series and a Strep-Tactin XT reference flow cell with no IL-2Rαβγ were used for double-referencing during data analysis. Raw data were fit to a 1:1 kinetic model with global analysis using Biacore Insight Evaluation software. All reported values were rounded to 1 decimal place.

STAT5 signaling experiments

To assess STAT5 signaling of naive human and cynomolgus monkey PBMCs, previously isolated PBMCs from normal donors were thawed and washed once with R10 medium. The cells were then resuspended in PBS containing Zombie NIR viability stain at 1:500 and incubated for 10 min at room temperature. The cells were then washed once with R10 medium and plated at 300,000 cells/well in warmed R10 medium in a round-bottom plate. Cells were incubated for 20 min at 37°C with 5% CO2 in R10 medium with a 12-point, dose titration from 100 nM to 6.1 pM in 1:4 dilutions and from 6.1 pM to 6.1 fM in 1:10 dilutions of either DF6215, aldesleukin, or non-alpha IL-2-Fc. To each well, 100 μL of 1X Working TFP Fix/Perm Buffer was then added, and the plate was incubated at 4°C for 50 min. Cells were washed twice with Perm/Wash Buffer and resuspended in 200 μL/well of ice-cold Perm Buffer III. The plate was set at −20°C overnight, washed twice with Perm/Wash Buffer, and blocked with 40 μL of Blocking Buffer for 10 min at 4°C. Subsequently, cells were stained with antibodies by adding 10 μL of 5X antibody cocktail (fluorochrome-conjugated anti-human or anti-cynomolgus monkey antibodies) and incubated for 40 min at 4°C in the dark. Cells were washed twice with Perm/Wash Buffer and resuspended in 100 μL of FACS buffer until acquisition on the Cytek Aurora Full Spectrum Cytometry System. All data from Cytek Aurora was analyzed using FlowJo. Within each cell subset analyzed (Tregs, CD8⁺ T cells, CD4⁺ T cells, and NK cells), the % positive pSTAT5 population for each concentration was plotted against the logarithm of the test article concentration, and the EC50 and EC90 of signaling was calculated based on a 4-parameter nonlinear regression analysis using GraphPad Prism v9.

To assess STAT5 signaling of TCR stimulated human and cynomolgus monkey PBMCs, previously isolated PBMCs were thawed, washed once with R10 media and resuspended in R10 medium to a concentration of 2.5 × 10⁶ cells/mL, and 25 μL of CD3/CD28 Dynabeads were added per 1 × 10⁶ cells. This mixture was incubated for 3 days, then the cells were separated from the magnetic beads using the EasySep Magnet. To separate the cells, the suspension was mixed vigorously to detach the cells from the beads, pipetted into a 14 mL polystyrene round-bottom tube, placed into the magnet for 1 min, and the supernatant cell suspension was poured off into a 50 mL conical tube. This cell suspension was spun down, resuspended in R10 medium, and allowed to rest at 37°C overnight for 16 h prior to stimulation with IL-2 variants and subsequent downstream assessment of STAT5 phosphorylation by flow cytometry.

Syngeneic tumor inoculation and treatment for efficacy/PD studies

Tumor cells were cultured and expanded for 7–9 days (3–4 passages). On the day of tumor inoculation, cells were harvested using TrypLE Express and washed twice in PBS. Cells were filtered through a sterile 70 μm strainer and counted. Subsequently, cells were adjusted to 2 × 10⁶ cells/mL or 5 × 10⁶ cells/mL for B16F10 and CT26, respectively, in sterile PBS. One hundred microliters of cell suspension were injected subcutaneously into the shaved right flank of mice using a 27-gauge needle/syringe. Subcutaneous tumor volumes were assessed by measuring the length, width, and height using a caliper, and tumor volume (mm³) was calculated as length x width x height. When tumors reached 50–200 mm³, mice were randomized into treatment groups. DF6215, non-alpha IL-2-Fc and isotype concentrations used were molar equivalent to rhIL-2 and dosed once weekly IP for a total of 4 doses. For combination studies with PD-1 blockade, anti-PD-1 (clone RMP1-14, BioXcell) was dosed twice weekly at 200 μg/mouse for 4 weeks.

Tissue processing for flow cytometry of mouse PD studies

Tumor tissues were harvested and weighed for endpoint measurement. Briefly, CT26 tumor tissue was diced into small pieces with scissors and placed into a Miltenyi C-tube containing an enzyme digestion mix according to the Miltenyi Tumor Dissociation Kit (Cat# 130-096-730). Tumor tissue was processed according to the Miltenyi Tumor Dissociation kit protocol. Once the cell suspensions were generated, 20 μL of count beads (Biolegend; Cat# 424902) were added and the samples were washed with PBS for downstream flow cytometry staining.

Flow cytometry

Whole blood or digested cell suspensions containing count beads were spun down and resuspended to 100 μL in staining buffer and then were added to a 96-well round-bottom plate. Samples were incubated with 0.5 mg/test Fc blocking antibody (BioLegend) and Zombie Near IR live/dead stain (BioLegend) for 10 min at room temperature in the dark. Subsequently, cells were washed and stained with antibodies against CD45 (clone 30-F11), CD3 (clone 17A2), CD8 (clone 53–6.7), CD4 (BD, clone GK1.5), CD49b (BD, clone HMα2), CD11b (clone M1/70), PD-1 (Invitrogen, clone J43), CD69 (eBioscience, clone H1.2F3), CD25 (BD, clone PC61), CD19 (clone 6D5), Ki-67 (BD, clone B56), CD62L (BD, clone MEL-14), CD44 (BD, clone IM7), Granzyme B (clone GB11), NKG2D (clone CX5), FOXP3 (eBioscience, clone FJK-16s) and TIM3 (clone RMT3-2.3) for 30 min at 4°C. All antibodies were purchased from BioLegend unless otherwise indicated. Intracellular staining and fixation were done using the eBioscience Foxp3/Transcription Factor Buffer set (Cat# 00-5523-00). Cells were washed and fixed according to buffer set instructions for future analysis with the Cytek Aurora Spectral Cytometer.

Flow cytometry of peripheral blood in naive mice

Naive BALB/c mice (6–8 weeks of age) were sorted into treatment groups and DF6215 or isotype was administered intraperitoneally (IP) according to study design. Mice were bled at specific timepoints and 50 μL of whole blood was used for flow cytometry staining. Fifty microliters of whole blood were transferred to a 96-well round-bottom plate and incubated with 150 μL of ACK lysis buffer for 5 min. Samples were quenched with 150 μL of FACS buffer and spun down. Samples were washed and spun 3 more times. Samples were then incubated with 0.5 mg/test Fc blocking antibody (BioLegend) and Zombie Near IR live/dead stain (BioLegend) for 10 min at room temperature in the dark. Subsequently, cells were washed and stained with antibodies against CD45 (clone 30-F11), CD3 (clone 17A2), CD8 (clone 53–6.7), CD4 (BD, clone GK1.5), CD49b (BD, clone HMα2), CD25 (BD, clone PC61), Ki-67 (BD, clone B56), FOXP3 (eBioscience, clone FJK-16s) for 30 min at 4°C. All antibodies were purchased from BioLegend unless otherwise indicated. Intracellular staining and fixation were done using the eBioscience Foxp3/Transcription Factor Buffer set (Cat# 00-5523-00). Cells were washed and fixed according to buffer set instructions for future analysis with the Cytek Aurora Spectral Cytometer.

Cytokine & chemokine response in tumor-bearing mice

CT26-Tyrp1 tumor-bearing BALB/c mice were randomized and equally distributed into treatment groups (n = 10/group) when tumors reached ∼190 mm³ and treated intraperitoneally (IP) with 0.675 or 0.225 mg/kg or human immunoglobulin isotype control (hIgG1 control). As a positive reference control, serum was collected from a mouse 2 h after treatment with anti-CD3 antibody. Mice were euthanized 24 h after treatment, and serum and tumors were collected and processed for cytokine analysis.

Tumor lysates were processed using the Bio-Plex Cell Lysis Kit (Biorad, Cat#171304011) according to manufacturer’s instructions. To normalize total protein concentration in the tumor lysates, a Peirce BCA Protein assay (Cat# 23225) was performed according to manufacturer instructions. Samples were normalized to 1 mg/mL total protein concentration prior to evaluation in the Luminex bead-based assay.

A bead-based Luminex platform-based 26-plex assay (Cytokine & Chemokine Convenience 26-Plex Mouse ProcartaPlex Panel 1 Cat# EPXR260-26088-901) was used for the detection of 26 cytokines in both serum and normalized tumor lysate samples. The protocol from the Cytokine & Chemokine Convenience 26-Plex Mouse ProcartaPlex Panel 1 User Guide was followed to prepare and run all samples on the Luminex xMAP Intelliflex. Cytokine data were analyzed using the ProcartaPlex Analysis software by Thermo Fisher Scientific.

Human PBMC cytokine release assay

Previously isolated human PBMCs from 5 healthy donors were thawed and resuspended at 1.5 × 10⁶ in complete R10 media (RPMI 1640, 10% HI-FBS, 1% penicillin/streptomycin, 1% GlutaMAX) and transferred into the appropriate wells of a 96-well round-bottom plate. Stimulants were added directly into the PBMC suspension and incubated in a humidified incubator at 37°C, 5% CO₂ for 48 h. Equivalent amounts of each compound in terms of μg/well were added and stimulation volume was kept constant during setup of the assay. All samples from the 5 healthy human donors were tested in triplicate for each assay condition. After the stimulation, assay plates were spun down, and cell culture supernatants were harvested and transferred to a new 96-well round-bottom microplate and stored at −80C until cytokine analysis by Luminex.

Toxicology study of DF6215 in cynomolgus monkey

Naive cynomolgus macaques were dosed by IV (slow bolus) injection with 30, 100 or 300 μg/kg of DF6215 on Day 1, 15, and 29 (n = 2–5/sex/group). Mortality, clinical signs, body weights, food consumption, and clinical pathology were monitored prior to and throughout the study. Following terminal and recovery sacrifice, gross necropsy, measurement of organ weights and histopathologic examinations were conducted.

Pharmacokinetic analysis of DF6215 in cynomolgus monkey

Blood samples were collected at pre-dose and 0.083, 2, 8, 24, 48, 72, and 168 h after dose administration. Samples were processed to serum, and the concentration of DF6215 was quantitated using a validated sandwich ELISA.

Pharmacodynamic analysis of DF6215 in cynomolgus monkey

Complete blood count (CBC) analysis was performed on cynomolgus monkey blood samples collected at baseline and post-treatment timepoints to monitor leukocyte differentials, hemoglobin, hematocrit, platelet count, and red blood cell indices. In parallel, immunophenotyping was conducted via flow cytometry to characterize changes in peripheral immune cell populations. In addition, plasma was collected to analyzed for cytokine levels. The plasma samples were analyzed for IL-1β, IL-1RA, IL-2, IL-5, IL-6, IL-8, IL-10, IL-12/23 (p40), IL-13, IL-17A, MCP-1, MIP-1β, IFN-γ, TNF-α, and GM-CSF using a validated bead-based multiplex immunoassay and analyzed by the Luminex detection system.

Quantification and statistical analysis

Data are presented as the mean ± SEM or the mean ± SD for representative data. All statistical analyses were performed using GraphPad Prism V8/9.0 software (GraphPad). Phoenix WinNonlin V8.1 software was used for assessment of PK parameters. One-way ANOVA or two-way ANOVA, unpaired t test and Tukey’s multiple comparison test were used to determine significance. Kaplan-Meier survival plots were prepared and analyzed using the log rank Mantel-Cox test. All flow cytometry analysis was performed using the FlowJo program (Treestar). p values less than 0.05 were considered statistically significant.

Published: December 29, 2025