Coupling IL-2 with IL-10 to mitigate toxicity and enhance antitumor immunity

Summary

Wild-type interleukin (IL)-2 induces anti-tumor immunity and toxicity, predominated by vascular leak syndrome (VLS) leading to edema, hypotension, organ toxicity, and regulatory T cell (Treg) expansion. Efforts to uncouple IL-2 toxicity from its potency have failed in the clinic. We hypothesize that IL-2 toxicity is driven by cytokine release syndrome (CRS) followed by VLS and that coupling IL-2 with IL-10 will ameliorate toxicity. Our data, generated using human primary cells, mouse models, and non-human primates, suggest that coupling of these cytokines prevents toxicity while retaining cytotoxic T cell activation and limiting Treg expansion. In syngeneic murine tumor models, DK2¹⁰ epidermal growth factor receptor (EGFR), an IL-2/IL-10 fusion molecule targeted to EGFR via an anti-EGFR single-chain variable fragment (scFV), potently activates T cells and natural killer (NK) cells and elicits interferon (IFN)γ-dependent anti-tumor function without peripheral inflammatory toxicity or Treg accumulation. Therefore, combining IL-2 with IL-10 uncouples toxicity from immune activation, leading to a balanced and pleiotropic anti-tumor immune response.

Graphical abstract

Highlights

• •IL-2 drives highly toxic cytokine release syndrome (CRS) and expansion of Tregs
• •IL-10 suppresses both CRS inflammation and Treg expansion
• •Coupling IL-2 with an IL-10 mutein suppresses CRS while enhancing T cell function
• •Targeting the molecule to the tumor enhances potency

Ahn et al. characterize the function of an IL-2+IL-10 tumor-targeted fusion protein. This molecule uncouples IL-2 inflammatory toxicity from potent adaptive anti-tumor immune activation by virtue of IL-10’s natural anti-inflammatory and Treg-suppressing functions. Targeting the tumor cell surface enables retention in the tumor and enhances potency.

Introduction

Significant efforts have been made in the past 30 years to change the immunosuppressive tumor microenvironment (TME) with immune-stimulating, pro-inflammatory cytokines. Interleukin (IL)-2 is a potent immunomodulator of T and natural killer (NK) cells that has been shown to enrich cytotoxic T lymphocytes (CTLs) in tumors with established clinical efficacy, namely with aldesleukin.¹ Wild-type IL-2 (wtIL-2) and PEGylated interferon (IFN)α have been approved for use in certain indications^2,3; however, use is limited by toxicities and regulatory T cell (Treg) responses. The high-affinity trimeric IL-2 receptor (IL-2R), consisting of IL-2Rα (CD25), IL-2Rβ (CD122), and the common gamma chain (CD132), is constitutively expressed on immunosuppressive Tregs. The dimeric IL-2R, lacking IL-2Rα, has 10–100 times lower affinity and is present on resting memory and effector T cells as well as NK cells.⁴ It is these immunostimulatory effects on effector T cells and NK cells that made high-dose IL-2 a promising cancer therapy.

In an effort to dissociate IL-2 toxicity from anti-tumor function, variants with impeded IL-2Rα binding have been evaluated.^5,6,7,8 Unfortunately, their development has been halted predominantly due to unfavorable toxicity and insufficient clinical function.⁹ Wu et al. showed in an MC38 colon carcinoma model that an IL-2-Fc variant with reduced IL-2Rα binding had little to no therapeutic efficacy despite having equivalent signal transducer and activator of transcription 5 (STAT5) induction to a wtIL-2-Fc.¹⁰ These results suggest that, for effective IL-2 cancer treatment, IL-2Rα binding potential should be retained.

The goal of our study is to solve the predominant challenges of harnessing the therapeutic utility of wtIL-2. It has been suggested that IL-2 toxicities are caused, in part, by IL-5 induction leading to eosinophilia and eosinophil degranulation^11,12 or increased CD44 expression on NK cells and subsequent activation.¹³ These in turn cause vascular leak syndrome (VLS) or capillary leak syndrome (CLS), and ultimately edema.¹⁴ Contrary to previous publications, we hypothesize that IL-2-induced VLS is preceded by the induction of high levels of multiple cytokines, leading to cytokine release syndrome (CRS). Others have shown that suppressing tumor necrosis factor alpha (TNF-α) increased the maximum tolerated dose of IL-2 and reduced hypotension and organ dysfunction.¹⁵ Also, IL-6 suppression is used to treat CRS from T cell-engaging therapies, namely chimeric antigen receptor T cells (CAR-T) and T cell engagers.¹⁶ Furthermore, in mice, the amelioration of CRS and VLS from COVID infection requires the concomitant neutralization of IFNγ and TNF-α.¹⁷

Alternatively, we propose using IL-10’s pleiotropic, immunoregulatory functions to address these factors limiting broader IL-2 implementation. IL-10 blocks cytokine secretion (e.g., TNF-α) in lipopolysaccharide (LPS)-stimulated monocytes and/or macrophages.¹⁸ Additionally, PEGylated recombinant human IL-10 (PEG-rhIL-10) can limit IL-2- and transforming growth factor β (TGF-β)-induced Treg expansion.¹⁹ In the clinic, low-dose IL-10 induced remission in approximately 25% of patients with Crohn’s disease and suppressed TNF-α secretion, while higher doses (10–20 mg/kg) were less effective, possibly from IFNγ induction.²⁰ IL-10’s immunostimulatory function has now been illustratively described with studies using IL-10-expressing syngeneic murine tumors,²¹ treatment of tumor-bearing mice with IL-10,^22,23 and clinical studies showing a 20%–30% objective response rate (ORR) in patients with cancer administered PEG-rhIL-10.²⁴

Given that IL-10 simultaneously suppresses inflammation, limits Treg accumulation, and enhances CTL anti-tumor function, we hypothesized that in addition to limiting IL-2 toxicity, the combined cytokines would retain or exhibit enhanced adaptive immune activation while limiting Tregs. In this study, we provide confirmatory data for the four primary hypotheses: (1) IL-2-mediated VLS is preceded by CRS, (2) IL-2-mediated VLS is induced by CRS and can be balanced and reduced through combination with IL-10, (3) IL-10 limits IL-2-driven Treg accumulation, and (4) the combination of cytokines increases anti-tumor potency compared to IL-2 alone.

Results

IL-10 controls IL-2-mediated inflammation and reduces toxicity

wtIL-2 is approved for renal cell carcinoma (RCC) and metastatic melanoma; however, associated toxicities such as CRS and VLS have limited its utility. We first sought to define the IL-2 inflammatory response in human peripheral blood mononuclear cells (PBMCs) and whether concomitant treatment with human IL-10 (hIL-10) could dissociate the secretion of CRS-contributing cytokines (i.e., IL-1β, IL-6, TNF-α) from critical T cell activation and induction of anti-tumor factors (i.e., IFNγ, perforin, and granzyme B [GzmB]). Both IL-2 and IL-2+IL-10 induced similar IFNγ secretion prior to T cell activation; however, IL-2+IL-10 resulted in significantly more perforin and GzmB (Figure 1A). Additionally, IL-2 alone induced TNF-α secretion, unlike IL-2+IL-10 (Figure 1B). Activation of T cells following IL-2 administration may contribute to CRS; therefore, we also measured these factors after anti-CD3. Both IL-2- (≥0.13 nM) and IL-2+IL-10 (≥1.3 nM)-treated cells showed dose-escalating increases in IFNγ, whereas IL-10 (≥0.13 nM) reduced IFNγ (Figure 1C). Additionally, both IL-2- and IL-2+IL-10-treated cells secreted more perforin (≥1.3 nM) and GzmB (≥0.13 nM) than controls. In keeping with IL-10’s known potentiation of IL-2 stimulation of T cells,²¹ IL-2+IL-10 resulted in more GzmB and perforin than IL-2 alone (Figure 1C). Notably, after anti-CD3, cells treated with IL-2 but not IL-2+IL-10 secreted TNF-α (IL-2 at 13 nM) and now IL-1β and IL-6 (both with IL-2 ≥ 1.3 nM) (Figure 1D). IL-10 alone tended to reduce TNF-α (p = 0.06 at 13 nM) with no consistent effect on IL-1β or IL-6.

Figure 1.

IL-10 controls IL-2-mediated induction of inflammatory cytokines

PBMCs were cultured with a titration of IL-2, IL-10, or IL-2+IL-10 (0–13 nM) for 24 h followed by measurement of supernatant analytes either before (A and B) or following anti-CD3 (C and D). Statistical analyses are depicted between IL-2 and IL-2+IL-10 groups at the highest concentration (13 nM) by donor-matched pairwise t test (∗p ≤ 0.05, ∗∗p ≤ 0.01, ∗∗∗p ≤ 0.001, ∗∗∗∗p ≤ 0.0001). Results are reported as the mean ± SEM.

To evaluate whether this is recapitulated in vivo, we dosed mice with IL-2, IL-10, or both concomitantly and evaluated serum markers of peripheral inflammation. Like the in vitro findings when activating T cells, mice treated with IL-2+IL-10 had markedly lower IFNγ, TNF-α, and IL-6 but similar GzmB compared to mice treated with IL-2 alone (Figures 2A–2D). Lower serum cytokines correlated with less edema, 0.03176 μg/lung (IL-2 alone) vs. 0.02298 μg/lung (IL-2+IL-10) of Evans blue, a reduction of 35.8% ± 14.5% when normalized to the control group (Figure 2E). Eosinophilia has also been suggested to contribute to IL-2 toxicity; as such, we measured eosinophil frequencies (SSC^hi/CD11b⁺/SiglecF⁺). IL-2 treatment resulted in a 3.2-fold increase in eosinophils among splenocytes compared to 2-fold with concomitant treatment (Figures 2F and S1A). Despite concomitantly treated animals having less inflammatory cytokines, reduced edema, and fewer eosinophils, these animals had distinct indications of immune activation. GzmB and perforin were increased in NK cells (NKp46⁺/GzmB⁺/Perforin⁺), equivalent between the IL-2-treated and concomitantly treated animals (Figures 2G and S1B). Additionally, both groups had a 3.6-fold increase in CD8⁺ effector memory T (Tem) cell (CD8⁺/CD44⁺/CD62L⁻) frequency among total spleen CD8⁺ T cells (Figures 2H and S1B).

Figure 2.

Concomitant treatment of IL-2 with IL-10 prevents IL-2-induced systemic inflammation and edema without stunting immune activation

(A–D) Mice were subcutaneously administered vehicle control, IL-2 (5 μg), IL-10 (2.4 μg), or IL-2+IL-10 (5 and 2.5 μg, respectively) daily for 9 days, as indicated. Statistical analyses comparing the control, IL-2, and IL-2+IL-10 groups are reported. (A–D) Serum analytes on day 9 after final administration were measured. Data are representative of 3 animals per group. Statistical analyses performed by two-way ANOVA with Tukey’s multiple comparisons test (∗∗p ≤ 0.01; ∗∗∗p ≤ 0.001; ∗∗∗∗p ≤ 0.0001).

(E) After final administration, Evans blue dye extravasation into the lungs was measured. Data are representative of 3–9 animals per group. Statistical analyses were performed by one-way ANOVA with Tukey’s multiple comparisons test (∗∗p ≤ 0.001; ∗∗∗∗p ≤ 0.0001).

(F–J) The frequency of immune cell subsets in spleen, namely eosinophils (SSC^hi/CD11b⁺/SiglecF⁺), GzmB⁺/Perforin⁺ double-positive NK cells (CD3⁻/CD19⁻/NKp44⁺), CD8⁺ Tem (CD3⁺/CD8⁺/CD44⁺/CD62L⁻), and Tregs (CD3⁺/CD4⁺/CD25⁺/Foxp3⁺). Data are representative of 3 animals per group. Statistical analyses were performed by one-way ANOVA with Tukey’s multiple comparisons test (p value; ns, not significant; ∗p ≤ 0.05; ∗∗p ≤ 0.01; ∗∗∗p ≤ 0.001; ∗∗∗∗p ≤ 0.0001).

Results are reported as the mean ± SEM.

Treg expansion is also detrimental to IL-2 treatment efficacy. The Treg (CD4⁺/CD25⁺/FoxP3⁺) frequency among total spleen CD4⁺ T cells increased 3.6-fold with IL-2 and only 2.5-fold with concomitant treatment (Figure 2I). Furthermore, the CD8⁺ Tem-to-Treg ratio trended higher with concomitant treatment (p = 0.07), whereas the IL-2 group showed no trend (p = 0.6) compared to control animals (Figure 2J). Taken together, these in vivo data corroborate the in vitro studies and establish how combination of IL-2 and IL-10 may address some of the factors limiting wider implementation of IL-2 in the clinic.

IFNγ- and TNF-α-dependent IL-2 toxicity is ameliorated by IL-10

To further assess how IL-10 limits IL-2-induced inflammatory cytokines, a temporal analysis of cytokine secretion on human PBMCs was conducted. These findings suggested that IFNγ is secreted first, followed by TNF-α, and finally IL-1β and IL-6 (Figures S2A–S2D). Other studies have suggested a similar temporal expression of CRS-associated cytokines following treatment of patients with T cell-based therapies.^25,26 Given that IL-2 alone and IL-2+IL-10 induced similar IFNγ secretion in PBMCs prior to anti-CD3 (Figure 1A), we tested if TNF-α, IL-1β, and IL-6 inductions were IFNγ dependent. We also tested if IL-10 feedback limited secretion because elevated IL-10 often corresponds with the onset of CRS. To do so, cells were treated with IL-2 alone or in combination with antibodies blocking either the IFNγ receptor 1 (IFNγR1) or IL-10 receptor 1 alpha (IL-10RA). Donors with a 2-fold induction of IFNγ in response to IL-2 are included, about 20% of donors’ PBMCs tested. Blockade of IFNγR1 or IL-10RA had no consistent effect on IFNγ, and the depicted trend was driven by just 1 donor experiment (Figure 3A). Blocking IFNγR1 abolished TNF-α, IL-1β, and IL-6 induction, whereas anti-IL-10RA had no consistent effect (Figures 3B–3D). These data suggest that IL-2 induction of TNF-α, IL-1β, and IL-6 in human PBMCs is IFNγ dependent. Therefore, we tested whether IFNγ alone is sufficient to induce TNF-α, IL-1β, and IL-6 as well as whether IL-10 could block this. IFNγ induced TNF-α and IL-10 blocked this (Figure 3E), while there was no significant induction of IL-1β or IL-6 (Figures 3F and 3G). Given that these three cytokines are predominantly secreted by monocytes and macrophages, we evaluated if IL-10 alters IFNγ signaling in monocytes. IFNγ induced interferon regulatory factors (IRF) 1 and 8, as expected, and the addition of IL-10 suppressed their expression (Figures 3H and S2E–S2G). Interestingly, TNF-α^−/− but not IFNγ^−/− mice were resistant to IL-2-mediated edema and anti-TNF-α prevented IL-2-mediated edema (Figures 3I and 3J), suggesting a critical role for TNF-α. Analyses of plasma IL-6 showed that IFNγ^−/− mice had five times higher levels than TNF-α^−/− mice after IL-2 administration (Figures S2H and S2I).

Figure 3.

IFNγ- and TNF-α-dependent IL-2 toxicity is ameliorated by IL-10

(A–D) PBMCs were untreated (control) or cultured with cytokines (1.3 nM) and receptor-blocking antibodies as follows: IL-2, IL-10, IL-2+anti-IFNγR1 (5 μg/mL), or IL-2+anti-IL-10RA (30 μg/mL) for up to 48 h, and supernatant analytes were measured. Data are representative of 5–9 independent donor experiments. Statistical analyses performed by donor pairwise t test on samples collected at 48 h. Statistically significant differences reported between IL-2, IL-2+anti-IFNγR1, and IL-2+anti-IL-10RA (p value; ∗p ≤ 0.05; ∗∗p ≤ 0.01; ∗∗∗p ≤ 0.001).

(E–G) PBMCs were cultured with IL-10 (1.3 nM), IFNγ (10 ng/mL), and IL-10+IFNγ or untreated (control), and supernatant analytes were measured. Data are representative of 4 independent donor experiments. Statistical analyses of the 48 h results were performed by pairwise t test (p value; ns, not significant; ∗∗p ≤ 0.01; ∗∗∗p ≤ 0.001).

(H) Monocytes were stimulated for up to 16 h with IFNγ alone or with IL-10 (1.3 nM) and stained for IRF1 and IRF8 protein levels. Reported is the median fluorescence intensity (MFI) of IFNγ+IL-10-treated cells relative to the IFNγ-treated cells. Data are representative of 3 independent donor experiments. Statistical analyses performed by one-sample t test (ns, not significant; ∗p ≤ 0.05; ∗∗p ≤ 0.01).

(I and J) Quantitation of Evans blue dye extravasation into the lungs after IL-2 treatment of wild-type, IFNγ-knockout, and TNF-α-knockout animals (I) and wild-type mice administered anti-TNF-α or an isotype control (J). Data are representative of 6–9 animals per group.

Statistical analysis by one-way ANOVA with Tukey’s multiple comparisons test (ns, not significant; ∗p ≤ 0.05; ∗∗p ≤ 0.01; ∗∗∗p ≤ 0.001; ∗∗∗∗p ≤ 0.0001).

Results are reported as the mean ± SEM.

Collectively, the in vitro findings suggest IL-10 can block IL-2-mediated induction of CRS-associated cytokines and that this may involve its ability to alter the response to IFNγ. The in vivo findings suggest a role of TNF-α in edema as well. Together, these suggest that by altering IFNγ and/or TNF-α signaling downstream of the IL-2 response, possibly with concomitant treatment of IL-10, CRS and edema may be mitigated. We therefore sought to design a fusion molecule combining IL-2 and IL-10 and to test if the toxicological effects of IL-2 are attenuated while retaining anti-tumor efficacy.

Structural combination of wtIL-2 with an IL-10 mutein for tumor-targeted delivery

A fusion protein of wtIL-2 and a mutein Epstein-Barr virus (EBV) IL-10 was engineered. Firstly, at 0.5–5 nM, hIL-10 and EBV IL-10 similarly suppressed TNF-α from LPS-stimulated human monocytes (Figure 4A). Conversely, human CTLs secreted more IFNγ with hIL-10 than with EBV IL-10, even at the highest concentrations tested (Figure 4B). This bifurcation of function was also established previously with muteins of hIL-10 and EBV IL-10.^27,28 We evaluated molecules containing EBV IL-10 with point mutations described by Yoon et al. to increase IL-10RA affinity²⁹ and coupled these to an anti-Ebola human antibody single-chain variable fragment (scFv) scaffold. This was because we intended for the final fusion molecule to have an scFv conferring tumor-associated antigen (TAA) targeting. These molecules, deemed scFv:EBV (scFv:EBV) mutein 1 (M1) and mutein 2 (M2), were tested alongside an uncoupled hIL-10 and an scFv:EBV with wild-type EBV IL-10. ScFv:EBV M2 has two point mutations of V31L and A75I compared to the single mutation in scFv:EBV M1. ScFv:EBV M2 suppressed TNF-α from LPS-treated monocytes and also induced IFNγ secretion from CTLs, comparable to hIL-10 (Figures 4C and 4D). Therefore, the DK2¹⁰ (EGFR) molecule was designed with the high-affinity mutein EBV IL-10 sequence from scFv:EBV M2 and wtIL-2 for conferring maximal suppression of any deleterious inflammatory response alongside maximal stimulation of T cells to synergize with IL-2’s anti-tumor activity. Additionally, complementarity determining region sequences from cetuximab, an anti-human epidermal growth factor receptor (EGFR) antibody, were engrafted into the scFv to confer TAA targeting. Several possible configurations for this final molecule were tested (Figures S3A–S3F); however, the most stable structure was achieved by linking IL-2 between the variable domains of the heavy (VH) and light chains (VL) of the scFv flanked by IL-10 monomers as shown in Figure S3B, with the sequence provided in Figure S3G. A model of the predicted crystal structure is depicted in Figure 4F.

Figure 4.

Structural combination of wild-type IL-2 and an IL-10 mutein

(A) Monocytes treated with LPS alone (controls) or in addition to a titration (0–13 nM) of wild-type or EBV IL-10. TNF-α reduction reported after 24 h compared to controls. Data are representative of 4 independent donor experiments. Statistical analyses performed by two-way ANOVA with Tukey’s multiple comparisons test and concentrations with statistical differences are indicated (∗p ≤ 0.05; ∗∗∗∗p ≤ 0.0001).

(B) CTLs (CD8⁺) treated with a titration (0–13 nM) of wild-type or EBV IL-10 (0–13 nM) and IFNγ measured after 24 h. Data are representative of 14 independent donor experiments. Statistical analyses performed by two-way ANOVA with Tukey’s multiple comparisons test (∗∗p ≤ 0.01).

(C) Monocytes treated with LPS alone (controls) or in addition to a titration (0–13 nM) of ScFV:EBV IL-10 wild type, ScFv:EBV IL-10 muteins (M1 or M2), or recombinant human IL-10 and TNF-α measured after 24 h. Data are reported as TNF-α reduction compared to controls. Data are representative of 2–4 independent donor experiments.

(D) CTLs (CD8⁺) treated with a titration (0–13 nM) of ScFV:EBV IL-10 wild type, ScFv:EBV IL-10 muteins (M1 or M2), or recombinant human IL-10 and IFNγ measured after 24 h. Data are representative of 6 independent donor experiments. Statistical analyses performed by one-way ANOVA at the highest concentration (∗∗p ≤ 0.01).

(E) Schematic of the full-length DK2¹⁰ (EGFR) construction depicting each domain.

(F) A predicted computational model of DK2¹⁰ (EGFR) generated using chimera, with domains indicated.

(G and H) The far- (G) and near-ultraviolet (H) spectra of DK2¹⁰ (EGFR) acquired via circular dichroism.

(I–K) The binding kinetics of DK2¹⁰ (EGFR) measured by Octet using streptavidin (SA) immobilized biotinylated EGFR (I), IL-2Rα (J), and IL-10RA (K).

Results are reported as the mean ± SEM.

Structural analysis showed DK2¹⁰ (EGFR) to be stably folded. Circular dichroism indicates domains present from each constituent part (i.e., wtIL-2, mutein EBV IL-10, and EGFR-directed scFv) (Figures 4G and 4H). Sedimentation velocity analytical ultracentrifugation analysis showed an expected sedimentation coefficient (S) of 3.5 with monomers (100%) having a frictional ratio of 1.34 (Figure S3H), suggesting an ellipsoidal hydrodynamic shape like that of therapeutic antibodies. Additionally, differential scanning fluorimetry indicated a first melting temperature after 66°C (Figure S3I). These data suggested that folding would allow each domain to interact with its cognate receptors. Indeed, the dissociation constant (K_D) of EGFR is K_D:10⁻⁹ (Figure 4I), approximately ½ that of cetuximab,³⁰ for IL-2Rα is K_D:10⁻⁹ (Figure 4J), and for IL-10RA is K_D:10⁻¹¹ (Figure 4K). The affinities of wtIL-2 (K_D:10⁻⁹) and hIL-10 (K_D:10⁻¹²) were similar (Figures S3J and S3K). Protein stability was elucidated through 24 months, at 5°C ± 3°C (Table S1). Taken together, these data suggest DK2¹⁰ (EGFR) is stable, functional, and has sufficient binding affinity for the cognate receptor of each domain.

DK2¹⁰ (EGFR) dissociates T cell immunostimulatory function from induction of CRS cytokines

To establish that DK2¹⁰ (EGFR) retained the immunostimulatory effects of wtIL-2 and IL-10 but dissociated the CRS-inducing cytokines, we first tested DK2¹⁰ (EGFR) in human primary cell monocultures. The molecule had an IC₅₀ of less than 0.01 nM (1 ng/mL) on LPS-stimulated monocytes (Figure 5A). Additionally, DK2¹⁰ (EGFR) primed T cells to secrete IFNγ (Figures 5B and 5C). Next, DK2¹⁰ (EGFR) was tested on human PBMCs. In PBMCs, DK2¹⁰ (EGFR) primed T cells to respond to T cell receptor (TCR) stimulation, evident from increased IFNγ, GzmB, and perforin secretion (Figure 5D), without TNF-α, IL-1β, or IL-6 (Figure 5E). Furthermore, DK2¹⁰ (EGFR) inhibited IRF1 and IRF8 expression on IFNγ-stimulated monocytes (Figure 5F), as was shown for IL-10 alone (Figure 3H). Taken together, these studies on human PBMCs established similar patterns of response to DK2¹⁰ (EGFR) and to wtIL-2 and hIL-10 concomitant treatment.

Figure 5.

Coupling IL-2 and IL-10 together retains immune activation and enhances tumor cytolysis

(A) Monocytes were treated with LPS alone (controls) or also with a titration (0–1.3 nM) of IL-2, IL-10, IL-2+IL-10, or DK2¹⁰ (EGFR) and TNF-α measured after 24 h. Reported as TNF-α reduction compared to controls and representative of 8 independent donor experiments. Statistical analyses indicated between IL-2, IL-2+IL-10, and DK2¹⁰ (EGFR) performed by two-way ANOVA with Tukey’s multiple comparisons test (∗p ≤ 0.05).

(B and C) CTLs (CD8⁺) (B) or CD4⁺ T cells (C) were treated with a titration (0–1.3 nM) of IL-2, IL-10, IL-2+IL-10, or DK2¹⁰ (EGFR) and IFNγ measured after 24 h. Data are representative of 8 independent donor experiments. Statistical analyses indicated between IL-2, IL-2+IL-10, and DK2¹⁰ (EGFR) performed by two-way ANOVA with Tukey’s multiple comparisons test (∗p ≤ 0.05; ∗∗p ≤ 0.01; ∗∗∗p ≤ 0.001).

(D and E) PBMCs were treated with a titration (0–1.3 nM) of DK2¹⁰ (EGFR) for 24 h (pre-anti-CD3) or stimulated an additional 24 h with anti-CD3 (post-anti-CD3). Anti-tumor factors (D) or CRS-associated factors (E) were measured in supernatant. Data are representative of 13–14 independent donor experiments. Statistical analyses were conducted by mix-effects analysis comparing analyte level at each dose of DK2¹⁰ (EGFR) to the respective pre/post-anti-CD3 untreated control (ns, not significant; ∗p ≤ 0.05; ∗∗p ≤ 0.01; ∗∗∗p ≤ 0.001; ∗∗∗∗p ≤ 0.0001).

(F) IRF1/IRF8 protein measured in monocytes stimulated up to 16 h with IFNγ (10 ng/mL) alone (controls) or with DK2¹⁰ (EGFR) (1.3 nM), and median fluorescence intensity (MFI) relative to the controls is reported. Data are representative of 3 independent donor experiments. Statistical analyses performed by one-sample t test (ns, not significant; ∗p ≤ 0.05).

(G–J) CTLs (CD8⁺) co-cultured with SK-BR3-GFP at a 10:1 E-to-T ratio. As indicated, cells were left untreated (controls) or treated with 1.3 nM of IL-2, IL-10, IL-2+IL-10, or DK2¹⁰ (EGFR) for up to 5 cycles of cytolysis. (H–I) During cycle 4 and 5 of cytolysis, DK2¹⁰ (EGFR) was additionally removed. (J) DK2¹⁰ (EGFR)-mediated cytolysis was measured after MHC I was knocked down in SK-BR-3-GFP cells by siRNA or MHC I antibody blocking. Data are representative of 3 independent donor experiments. Statistical analyses were performed by one-way ANOVA of the results at the final time point measured (∗p ≤ 0.05; ∗∗p ≤ 0.01; ∗∗∗p ≤ 0.001; ∗∗∗∗p ≤ 0.0001).

Results are reported as the mean ± SEM.

DK2¹⁰ (EGFR) enhances human CTL cytotoxic function in vitro and limits Treg expansion

Given our findings and previous reports that IL-2 and IL-10 enhance CTL function^21,31 and that IL-10 can enhance T cell fitness, we next compared how IL-2, IL-10, both these, or DK2¹⁰ (EGFR) impacted these using a human breast cancer cell line, SK-BR-3-GFP. After one round of cytolysis, DK2¹⁰ (EGFR) elicited similar cytotoxic potency from CTLs as IL-2 alone (Figure 5G). Persistent CTL cytolytic function is crucial to immunotherapy durability; therefore, we performed 5 rounds of cytolysis re-exposing CTLs to fresh tumor cells, media, and cytokines every 5 days. Cytolytic capacity persisted for all 5 rounds with DK2¹⁰ (EGFR) but only when DK2¹⁰ (EGFR) remained present (Figures 5G–5I and S4A). Knockdown of major histocompatibility complex class I (MHC I) by small interfering RNA (siRNA) and blockade of MHC I with antibodies showed cytolysis to be MHC I restricted (Figure 5J). In addition, unlike some IL-2 muteins that attenuate NK cell activation,³² DK2¹⁰ (EGFR) exhibited similar effects on NK cell cytolysis as both IL-2 and IL-2+IL-10 (Figure S4B). Given the persistent cytolytic capacity of DK2¹⁰ (EGFR)-treated CTLs, we conducted metabolic analyses. The oxygen consumption rates (OCRs), extracellular acidification rates (ECARs), spare respiratory capacity, and adenosine triphosphate production of DK2¹⁰ (EGFR)-treated cells were enhanced compared to controls and IFNγ levels elevated, similarly to IL-2-treated cells (Figures S4C–S4F). Phenotypic analysis of proliferating CTLs showed a similar proportion of stem-like memory (CD45RO⁻/CD62L⁺/CD95⁺) and central memory T cells (CD45RO⁺/CD62L⁺) between IL-2 and DK2¹⁰ (EGFR) conditions but slightly smaller proportions of Tem (CD45RO⁺/CD62L⁻/Ki67⁺) and the more terminally differentiated effector memory cells re-expressing CD45RA (Temra, CD45RO⁻/CD62L⁻/CD27^high/low) among DK2¹⁰ (EGFR)-treated cells (Figures S4G–S4L). This differed from IL-2+IL-10 concomitant addition that closely mirrored IL-2 alone. The cytotoxic function of CTLs increases as they differentiate toward Temra, but their self-renewal capacity decreases.³³ A previous study showed that molar excess levels of IL-10 compared to IL-2 could prevent Treg expansion among CD4⁺ T cells.¹⁹ Culturing of T cells with TGF-β and IL-2 resulted in a greater frequency of Tregs (CD4⁺/CD25⁺/CD127^low/FoxP3⁺) among CD4⁺ T cells and a lower ratio of CTLs to Tregs than with TGF-β and DK2¹⁰ (EGFR) (Figures S4M–S4O).

Given the enhancement in cytolytic function of human CTLs in vitro and the potential to limit Treg expansion compared to IL-2, we next sought to investigate the anti-tumor potency of DK2¹⁰ (EGFR) in murine tumor models and to compare this to treatment with IL-2, as well as to assess if targeting these molecules to the TME could enhance efficacy.

Coupling and targeting IL-2 and IL-10 together retains broad immune activation and enhances anti-tumor immunity in mice

Secretion of cytokines in the TME induces anti-tumor responses in mice and humans.³⁴ We therefore designed a coupling system that enabled TME enrichment of linked wtIL-2 with a high-affinity mutein of EBV IL-10.²⁹ Evaluation of the anti-tumor efficacy of wtIL-2 and hIL-10 in the syngeneic murine B16F10 tumor model stably transfected to express the extracellular domain of human EGFR (B16F10^hEGFR+) suggests that concomitant treatment exhibited greater anti-tumor potency than either cytokine alone (Figures 6A and S5A). Also, coupling to an untargeted scFv enables half-life extension but was insufficient to enhance anti-tumor function (Figures 6B and S5B); however, an anti-human EGFR-scFv enhanced anti-tumor function of all three treatments (Figures 6C and S5C). The testing of DK2¹⁰ (EGFR) in this murine tumor model illustrated a dose-titratable anti-tumor response, reaching saturation at 2 mg/kg (Figure 6D). Furthermore, DK2¹⁰ (EGFR)-treated mice considered “cured” exhibited no tumor regrowth after 1 month without treatment and were completely resistant to tumor re-challenge (Figure S5D). As with EGFR-targeting single cytokines, DK2¹⁰ (EGFR) at 2 mg/kg exhibited increased anti-tumor function compared to an untargeted DK2¹⁰ (scFv) (Figure 6E). Biodistribution studies of DK2¹⁰ (EGFR) (2 mg/kg dose) showed that approximately 400 ng/mL (5.5 nM) was retained in the tumor for at least 3 days after a single dose (Figures S5E and S5F). These findings illustrate that the retention of cytokines within the TME enhances anti-tumor function.

Figure 6.

Coupling and targeting IL-2 and IL-10 together retains broad immune activation and increases anti-tumor potency in mice

(A–M) C57BL/6 mice (C57BL/6 and IFNγ-knockout mice in H) were implanted subcutaneously with a human EGFR-expressing B16F10 cell line (B16F10^hEGFR+). B16F10^hEGFR+ tumor-bearing mice were treated three times per week with vehicle control or the reagent(s) indicated. (A–F, H–L) Data are representative of the tumor volumes (mm³) from 4 to 15 animals per group.

(A) Mice treated with vehicle, IL-2 (0.4 mg/kg), IL-10 (1.0 mg/kg), or both (n = 10). Statistical analyses were performed by Mann-Whitney test.

(B) Mice treated with vehicle, IL-2 (scFv) (0.4 mg/kg), IL-10 (scFv) (1.0 mg/kg), or both (n = 10). Statistical analyses were performed by two-way ANOVA with Tukey’s multiple comparisons test.

(C) Mice treated with vehicle, IL-2 (EGFR) (0.4 mg/kg), IL-10 (EGFR) (1.0 mg/kg), or both (n = 10). Statistical analyses were performed by two-way ANOVA with Tukey’s multiple comparisons test.

(D) Mice treated with vehicle control or varying doses (0–6 mg/kg) of DK2¹⁰ (EGFR) (n = 7). Statistical analyses were performed by two-way ANOVA with Tukey’s multiple comparisons test.

(E) Mice treated with vehicle, DK2¹⁰ (scFv) (2.0 mg/kg), or DK2¹⁰ (EGFR) (2.0 mg/kg) (n = 9–10). Statistical analyses were performed by Mann-Whitney test.

(F) Mice treated with vehicle, DK2¹⁰ (EGFR) (2.0 mg/kg), or varying doses of IL-2 (0–4.0 mg/kg) (n = 4–7). Statistical analyses were performed by two-way ANOVA with Tukey’s multiple comparisons test.

(G) Mice treated with vehicle, DK2¹⁰ (EGFR) (2.0 mg/kg), or IL-2 (2.0 mg/kg) (n = 4–7). Serum was collected, and cytokine concentrations are reported. Statistical analyses were performed by Student’s t test. Results are reported as the mean ± SD.

(H) Wild-type or IFNγ-knockout mice were treated with vehicle or DK2¹⁰ (EGFR) (2.0 mg/kg) (n = 7–9). Statistical analyses were performed by Welch’s t test.

(I–K) Mice treated with vehicle or DK2¹⁰ (EGFR) (2.0 mg/kg) with and without cell ablation of CD8⁺ (I), CD4⁺ (J), or NK cells (K) (n = 5–6). Statistical analyses were performed by Mann-Whitney test.

(L) Mice treated with vehicle or DK2¹⁰ (EGFR) (2.0 mg/kg) with and without FTY720 (n = 11–15). Statistical analyses were performed by Mann-Whitney test.

(M–P) Mice were treated with vehicle or DK2¹⁰ (EGFR) (2.0 mg/kg) for 6–8 days (n = 4). (M) The frequencies of CD4⁺ and CD8⁺ T cells among TILs. (N) Following treatment, CD8⁺ Tem (CD3⁺/CD8⁺/CD44⁺/CD62L⁻) TILs were sorted, and gene expression was performed by nCounter analysis. Genes with a fold change of 1.5 from control animals and p value ≤0.05 are reported. (O) The oxygen consumption rate of isolated CD8⁺ TILs (n = 9). (P) Isolated CD8⁺ TILs from treated mice were rested overnight before co-culturing with B16F10^hEGFR+ cells, and an Elispot was performed to detect IFNγ-producing tumor-reactive cells (n = 5–9). Statistical analyses were performed by Welch’s t test.

(Q) Tumor volumes (mm³) from mice bearing LL2^hEGFR+ tumors treated with vehicle or DK2¹⁰ (EGFR) with or without anti-PD-1 (2.5 mg/dose) for 1 week (n = 9). Statistical analyses were performed by Welch’s t test.

(R) A patient-derived xenograft (PDX) model (non-small cell lung cancer) was established by subcutaneous implant of patient tumor cells, and mice were treated with vehicle, anti-PD-1 (10 mg/kg; every 5 days intraperitoneally), or DK2¹⁰ (EGFR) (0.5 or 2.0 mg/kg, three times a week subcutaneously) (n = 2). Tumor volumes are reported (mm³). Statistical analyses were performed by Welch’s t test.

(ns, not significant; p value; ∗p ≤ 0.05; ∗∗p ≤ 0.01; ∗∗∗p ≤ 0.001; ∗∗∗∗p ≤ 0.0001).

Results are reported as the mean ± SEM, unless otherwise noted.

Importantly, a 5-fold molar excess of wtIL-2 (2 mg/kg) was needed to achieve similar anti-tumor efficacy as 2 mg/kg DK2¹⁰ (EGFR) (Figure 6F). In addition to exhibiting greater anti-tumor potency than wtIL-2, DK2¹⁰ (EGFR)-treated mice had lower serum IFNγ and TNF-α but similar serum IL-5 and GzmB (Figure 6G). Despite the similar levels of IL-5, DK2¹⁰ (EGFR)-treated animals exhibited no edema (Figure S5G). These data are corroborated by qPCR profiling of the two functionally equivalent doses of DK2¹⁰ (EGFR) and wtIL-2 (Figure S5H), where only the IFNγ-induced chemokines Cxcl10 and Cxcl11 are significantly higher with wtIL-2. Similar to the effects shown earlier of IL-2+IL-10 on Treg frequencies in the spleen of mice and of DK2¹⁰ (EGFR) on human Treg expansion in vitro (Figures 2I and S4), the ratio of CD8⁺ T cell to Treg within the tumor was greater in the DK2¹⁰ (EGFR) treatment group than in the control (Figures S5I and S5K). These studies establish that DK2¹⁰ (EGFR) dissociates the anti-tumor effects of IL-2 from the subsequent systemic inflammation and edema, as with IL-2+IL-10. Assessment of animal body weights showed that the administration of DK2¹⁰ (EGFR) did not affect animal weights whereas the weights of animals receiving IL-2 trended higher, but there was no statistically significant difference nor obvious IL-2 dose-related trend (Figures S5L and S5M).

Previous murine studies have illustrated that the effectiveness of IL-2 therapies requires multiple components of the immune system. Similarly, testing of DK2¹⁰ (EGFR) potency showed that endogenous IFNγ, CD8⁺ T cells, CD4⁺ T cells, and NK cells are important (Figures 6H–6K), as treatment is less effective in IFNγ^−/− mice and in mice where any of these cell subsets were ablated. We additionally tested whether de novo immune infiltration was required for DK2¹⁰ (EGFR)’s anti-tumor response or whether T cells present within the TME were sufficient for anti-tumor function. Co-administration of mice with FTY720 (inhibitor of lymphocyte trafficking) prior to DK2¹⁰ (EGFR) resulted in similar anti-tumor responses (Figure 6L). These data suggest that DK2¹⁰ (EGFR) primes anti-tumor immunity from just the immunological infiltrate within the tumor. DK2¹⁰ (EGFR) treatment augmented the intratumoral CD4⁺ and CD8⁺ T cell ratio by increasing the frequency of CD8⁺ T cells among tumor-infiltrating lymphocytes (TILs) (Figure 6M). Additionally, the memory phenotype of T cells shifted from a predominantly central memory (CD4⁺ or CD8⁺/CD44⁺/CD62L⁺) subset to effector memory (CD4⁺ or CD8⁺/CD44⁺/CD62L⁻) subset with DK2¹⁰ (EGFR) (Figures S5N–S5O). Moreover, effector memory CD8⁺ T cells sorted from TILs exhibit broad changes to mRNA expression patterns that suggested heightened cytolytic capacity (Gzma, Gzmb, Tnfsf9, and Tnfsf10), greater activation (Ifit1, Ifit3, Ifitm2, Ifitm3, Rsad2, Cd69, and Tnfsf10), and a shift in metabolic function (Ier3, Sgk1, Bnip3, and Map3k8) (Figure 6N). Sorted CD8⁺ TILs exhibited enhanced OCR (Figure 6O), ECAR, spare respiratory capacity, and ATP production rate (Figures S5P–S5R). Also, these CD8⁺ cells exhibit increased tumor reactivity (Figure 6P) and greater perforin/GzmB double positivity and frequency of Ki-67^high proliferating cells (Figures S5S and S5T).

There is interest in combination therapy with IL-2 therapeutics, such as combining with checkpoint blockades, including anti-programmed cell death protein 1 (PD-1). Checkpoint blockades are clinically efficacious.³⁵ To interrogate how DK2¹⁰ (EGFR) can be used in conjunction with these therapies, we used a Lewis Lung (LL2^hEGFR+) syngeneic tumor model and found that DK2¹⁰ (EGFR) in combination with anti-PD-1 exhibited greater protection against tumor outgrowth and greater therapeutic efficacy than either treatment alone (Figures 6Q and S5U). Interestingly, in a PD-1-resistant, non-small cell lung cancer patient-derived xenograft in the context of a CD34⁺ stem cell-reconstituted human immune system model, DK2¹⁰ (EGFR) treatment exhibited differential anti-tumor function (Figure 6R) compared to PD-1.

Coupling and targeting IL-2 and IL-10 safely uncouples IL-2-associated toxicity from T cell activation in non-human primates

Given these in vitro and in vivo effects of DK2¹⁰ (EGFR), we conducted a Good Laboratory Practice (GLP) non-human primate (NHP) toxicology study in cynomolgus monkeys by dosing three times a week subcutaneously at 0.25, 1.0, and 2.5 mg/kg. Cetuximab is cross-reactive to the EGFR of cynomolgus monkey.^36,37 The doses were selected based on the effective concentration 90 (EC₉₀) for human T cell stimulation in vitro at 1–10 ng/mL (Figure S6A) and previous toxicological studies in NHP conducted with IL-10³⁸ and DK2¹⁰ (EGFR). Pharmacokinetic evaluation showed that all three dosing levels achieved plasma concentrations at or above the EC₉₀, between 50 and 2,000 ng/mL at Cmax (Figure 7A). Consistent with the hypothesis that concomitant treatment of IL-2 with IL-10 would not result in CRS or subsequent VLS, none of the study subjects exhibited an elevation of plasma cytokines indicative of CRS (Figures 7B–7E). The ranges of IFNγ, TNF-α, and IL-1β induced in patients with cancer treated with high-dose IL-2 have been previously reported.^39,40 A maximally tolerated dose of 2.5 mg/kg and no observable adverse effect level of 1 mg/kg indicates a potential 10× safety window in excess of the projected human equivalent therapeutic dose (HED) of 0.25 mg/kg. Similar to IL-2, treatment with DK2¹⁰ (EGFR) increases peripheral blood eosinophils, but unlike IL-10, it does not reduce platelets or red blood cells at these doses and does not change total peripheral white blood cells (Figure 7F). Despite the 40-fold or greater increases in eosinophils, there were no signs of liver toxicities indicated by changes in alanine transaminase, aspartate transaminase, or gamma-glutamyl transferase and no signs of anemia indicated by changes to hemoglobin nor red blood cell distribution width, nor lowering of cholesterol (Figures 7F and 7G). Additionally, no dose related-changes in weight were observed in either male or female animals (Figures S6B and S6C). Dose-dependent frequency changes in Tregs were observed at the highest dose, as well as proliferating cells of Tregs, CD4⁺ T cell, and cytotoxic CD8⁺ T cell subsets (Figures 7H–7K). However, the only time point and dose group at which the ratio of CD8⁺ T cells to Tregs differed from the corresponding controls was the 1.0 mg/kg group on day 27, and 72 h after administration on that day, this dose is 4× the projected HED, and the higher 2.5 mg/kg dose did not show a difference (Figure 7L). The markers used for immunophenotyping populations in the NHP study are provided in Table S2. Collectively, these results demonstrate the in vivo safety of administering DK2¹⁰ (EGFR).

Figure 7.

Coupling and targeting of IL-2 and IL-10 exerts broad immune activation in non-human primates and shows a strong safety profile in a GLP study

(A–L) Male, and female, cynomolgus monkeys were treated with DK2¹⁰ (EGFR) (0–2.5 mg/kg) dosed subcutaneously three times a week for 27 days. Each group had both males and females (n = 6–10).

(A) DK2¹⁰ (EGFR) plasma concentration plotted up to 72 h after the first dose.

(B–E) Plasma cytokines (pg/mL) from animals at the indicated study time points. Concentrations below the lower limit of quantification (LLOQ) are reported as 150 pg/mL.

(F) Hematology analyses, represented as fold change, comparing week −1 (pre-dose) to day 28 (post-dose) cell concentrations. Reported are eosinophils, platelets, red blood cells (RBCs), and white blood cells (WBCs).

(G) Clinical chemistry on day 28. Reported are alanine transaminase (ALT), aspartate aminotransferase (AST), gamma-glutamyl transferase (GGT), hemoglobin (HGB), red blood cell distribution width (RDW), and cholesterol.

(H–L) Blood was collected at the indicated times (time from first dose and the most recent dose are indicated on the x axis) to determine immune cell subset frequencies. Subsets reported are total T cells among CD45⁺ lymphocytes (H), frequency of proliferating (Ki67⁺) Tregs (CD3⁺/CD4⁺/CD127^low/CD25⁺) (I), CD4⁺ T cells (J), CTLs (K), and the ratio of CTLs to Tregs (L). Statistical analyses were performed by one-way ANOVA with Tukey’s multiple comparisons test on measurement at the final time point.

Results are reported as the mean ± SD.

LLOQ is indicated on several plots.

Discussion

Significant knowledge has accrued in 30 years to elucidate what is needed for effective, therapeutically useful immunotherapies. Though approved and still used in a limited fashion, high-dose IL-2 leads to sometimes lethal, toxicological burden.⁴¹ Engineered IL-2 variants with preferential dimeric IL-2Rβ/γ activity have been largely discontinued in clinical trials.^42,43 Therefore, engaging IL-2Rα may be critical to potency, as was reported in a murine MC38 tumor model.¹⁰ An approach to mitigating CRS following high-dose IL-2 and T cell-based therapies is to give anti-TNF-α or anti-IL-6/IL-6R. Alternatively, we propose a platform-based approach harnessing the immunoregulatory capacity of IL-10 to prevent toxicities and enhance anti-tumor potency. In this manuscript, we provided evidence for the following: (1) the toxicity of IL-2 is driven by uncontrolled CRS; (2) IL-2’s toxic inflammation can be ameliorated through combination with IL-10; (3) coupling IL-2 with IL-10 limits Treg accumulation; and lastly, (4) combination and targeting of the two cytokines to the TME enhances potency.

The toxicity of IL-2 has been attributed to IL-5 induction of eosinophilia and NK cell activation, triggering the cascade causing VLS or CLS, and ultimately edema and organ failure.^11,12,13 Contrary to this hypothesis, we interrogated whether a CRS cascade may precede edema using mouse models. Initial experiments demonstrated that PBMCs secreted CRS-associated cytokines including TNF-α, IL-1β, and IL-6 in response to IL-2, and IL-10 limited this (Figure 1). Also, this was dependent on a feedforward loop involving IFNγ (Figure 3). We showed that IL-10 alters IFNγ signaling in monocytes and inhibited IFNγ-induced IRF1 and IRF8 expression (Figure 3H), key transcription factors regulating TNF-α production.^44,45,46,47 Notably, IFNγ alone did not induce IL-1β and IL-6 within 48 h suggesting that additional factors are involved. Translating this in vivo, IFNγ^−/− mice presented with edema after high-dose IL-2, but TNF-α^−/− mice were resistant (Figure 3I). Synergistic activities of IFNγ and TNF-α can cause intestinal epithelial cell death,⁴⁸ and a SARS-CoV-2 study in mice showed that cytokine shock and organ failure were only blocked completely with concomitant anti-TNF-α and anti-IFNγ.¹⁷ IFNγ primes the inflammatory response of monocytes, and both TNF-α and IFNγ, but not either alone, were shown to induce IL-6.⁴⁹ In a clean in vitro system without serum, IL-1β and IL-6 following IL-2 exposure may require IFNγ. Whereas, in mice, additional environmental factors and low levels of inflammation may afford susceptibility to IL-2-induced CRS and edema without IFNγ. Further support for this comes from the fact that IFNγ^−/− mice had five times higher plasma IL-6 than TNF-α^−/− mice (Figure S2I). These data suggest that IL-2 toxicity is mediated by IFNγ and/or TNF-α and can be ameliorated with IL-10. In vivo investigations further demonstrated the detoxifying potential of IL-10 without it shunting the expansion of CD8⁺ Tem or induction of cytolytic factors (GzmB/perforin) by NK cells (Figure 2).

Expansion of Tregs across a broad dose range of IL-2 is reported to hamper its anti-tumor effects.^50,51,52 In CD4⁺ T cells, TGF-β induces FoxP3⁺ CD4⁺ Tregs and IL-2 stabilizes FoxP3⁺ expression.^53,54,55,56 IL-10 can regulate TGF-β-mediated effects,^57,58 and PEGylated IL-10 was reported to suppress intratumoral FoxP3⁺ Tregs.¹⁹ In this report, PEG-rhIL-10 inhibited IL-2+TGF-β-mediated Treg polarization of human CD4⁺ T cells. Though controversial at the time, PEG-rhIL-10 treatment resulted in immune activation and CD8⁺ T cell compartment remodeling to induce 25% monotherapy ORR in RCC and suppression of Treg accumulation.⁵⁹ Additionally, PEG-rhIL-10 led to little change in peripheral Tregs.²⁴ Administration of IL-10 with IL-2 limited Treg expansion in mice and showed a trending increase in the CD8⁺ T cell to Treg ratio compared to IL-2 alone (Figures 2I and 2J). Additionally, tumor-bearing mice treated with DK2¹⁰ (EGFR) had a greater CD8⁺ T cell to Treg ratio among TILs than the control group (Figure S5J). Also, in human T cells expanded under a Treg-polarizing condition (IL-2+TGF-β), the CD8⁺ T cell to Treg ratio was equivalent or less than TGF-β-treated cells, suggesting that IL-2 indeed favored Treg expansion, whereas DK2¹⁰ (EGFR) resulted in a small but consistent increase in the relative frequency of CD8⁺ T cells (Figure S4N).

Lastly, we investigated how the anti-tumor potency of IL-2+IL-10 or DK2¹⁰ (EGFR) compares to IL-2 alone. Targeted DK2¹⁰ (EGFR) was enriched within the hEGFR-transgenic B16F10 tumor of mice compared to the non-targeted molecule (Figure S5F). Tumor enrichment improved potency, and TAA targeting was required for curative function at the doses tested (Figures 6E and S5). Curative function with IL-2 required a 4- to 5-fold greater molar equivalent dose than DK2¹⁰ (EGFR). IL-10 is reported to prevent cytolytic dysfunction and rescue TILs from exhaustion, hence the interest in IL-10-producing CAR-T cells.^60,61 Notable is the observation that IL-10 was highly expressed in exhausted cells of lymphocytic choriomeningitis virus (LCMV)-infected mice,⁶² possibly as a mechanism to prevent dysfunction; however, that is speculative. The CTLs of these mice treated with wtIL-2 and anti-PD-1 expressed less IL-10 than in mice treated with anti-PD-1 and a reduced IL-2Rα affinity IL-2 variant.⁶³ Additionally, IL-2Rα engagement was important for mitigating viral load, and efficacy was associated with increased IL-2Rα and IL-18R1 and decreased CXCR5.⁶³ A similar pattern of expression was observed with DK2¹⁰ (EGFR)-treated mice compared to controls. In another study with OT-1 mice administered OVA-pulsed dendritic cells, the presence of Treg inhibited the expression of interferon-induced genes (Ifitm1 and Ifitm2), Gzma, Gzmb, and IL18R1 by CTLs.⁶⁴ A higher expression of these genes was observed in DK2¹⁰ (EGFR)-treated mice, while notably, also observing an increase in the CD8⁺ Tem to Treg ratio within the tumor. This increase was observed although systemic levels of IFNγ remain largely unchanged suggesting tumor proximal increases in the expression of IFNγ. This may contribute to a better safety profile than high-dose IL-2 and increase anti-tumor activity without edema.

The BCL-2 family of genes is important for cell survival, and Bcl2l1 was expressed more in CD8⁺ Tem of mice treated with DK2¹⁰ (EGFR) (Figure 6N). In CAR-T cells from patients with multiple myeloma that relapsed after BCMA CAR-T therapy, Bcl2l1 expression was lost, and Bcl2l1 armoring protected cells from activation-induced cell death and appeared to reduce functional exhaustion.⁶⁵ As CTLs differentiate, their stemness and ability to self-renewal progressively diminish, while their anti-tumor functions, namely cytotoxicity and IFNγ expression, increase.³³ Although the gene expression of human CTLs from cytotoxicity experiments was not measured, it is plausible that a similar expression pattern could have contributed to CTL potency being retained with DK2¹⁰ (EGFR) (Figure 5). These CTLs had increased OCR, ECAR, spare respiratory capacity, and ATP production comparable to the IL-2 condition (Figures S4C–S4E).

Immunotherapies including CAR-T cells, bispecific T cell engagers, engineered TCR-T cells, and adoptively transferred TILs, while effective in some patients, suffer from a limited therapeutic window caused in part by CRS and VLS.^{66,67,68,69,70} Moreover, CRS does not necessarily correlate with demonstrable anti-tumor function. The first cytokines secreted after activation of T cells are IL-2 and IFNγ.^71,72,73 Both cytokines induce further T cell activation (IL-2) and antigen presentation (IFNγ).^31,74 Our data suggest that IL-2 induces IFNγ while in turn IFNγ induces TNF-α, and TNF-α may be responsible for IL-2-mediated toxicity. IL-10’s control of IRF1/8 induction and blockade of TNF-α induction did not diminish immune activation. Therefore, these data suggest that the use of DK2¹⁰ (EGFR) with T cell-based therapies may enhance their effectiveness and suppress a toxicological CRS circuit.

With the forward momentum of immunotherapies in oncology experiencing a temporary lull when the scientific community has learned the most about how best to clinically harness the immune system, it behooves us to collectively apply our understanding through rational combinatorial approaches for the next generation of immunotherapies. Given this, we have initiated a series of preclinical investigations to ascertain how DK2¹⁰ (EGFR) can be utilized with the aforementioned modalities. Thus far, the observations presented in this study have been translated to patients with cancer in a phase 1 clinical trial (NCT05704985), suggesting that the combination of IL-2 with IL-10 detoxifies IL-2 and activates the immune system. Given the data presented in this report, we posit that DK2¹⁰ (EGFR) represents the beginning of the next generation of targeted approaches and will function as the basis for enhanced combinatorial therapies.

Limitations of the study

Our study primarily utilizes murine models with female animals, primary human PBMCs, and an NHP toxicological study using both male and female cynomolgus monkeys. Future combination trials with DK2¹⁰ (EGFR) are warranted, and a phase 1 clinical trial (NCT05704985) is ongoing to further substantiate the findings from this study.

Resource availability

Lead contact

Further technical or related inquiries and request for reagents may be directed to and will be fulfilled by the lead contact, John B. Mumm (mummj@dekabiosciences.com).

Materials availability

DK2¹⁰ (EGFR) generated in this study will be made available on request. Payment and/or a completed materials transfer agreement will be required if there is potential for commercial application.

Data and code availability

• •Nanostring nCounter data have been deposited at https://doi.org/10.17632/7skswgr2h5.1 and are publicly available as of the date of publication. Accession numbers are listed in the key resources table.
• •This paper does not report original code.
• •Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.

Acknowledgments

We thank Jeffrey Collins for the generation of ⁸⁹Zr DK2¹⁰ (EGFR) and Dr. Shili Xu and Mikayla Tamboline for conducting the in vivo biodistribution study at the Crump Institute’s Molecular Imaging Technology Center at University of California, Los Angeles. This work was funded by Deka Biosciences.

Author contributions

J.J.A., S.S.V.A., A.L.C., S.D., B.I.G., J.G., J.A.G.V., Y.M.H., F.W.H., J.E.J., P.N.K., D.P.L., J.C.M., R.R., A.S., J.D.S., S.R.S., Y.W., and A.W. performed the experiments and data analysis. D.P.L. conducted additional studies and the final manuscript preparation during revision. J.J.A., S.D., A.H.H., D.P.L., J.C.M., and S.R.S. contributed to the manuscript preparation. D.B., P.A.K., and C.H.Y. led the study. J.B.M. conceptualized and directed the study.

Declaration of interests

J.B.M., C.H.Y., and P.A.K. are current employees and shareholders at Deka Biosciences (“Deka”). Every author on this submission has been granted employee stock options as a part of their employment with Deka. J.B.M. is the president & CEO and board director at Deka. MPM BioImpact, Leaps by Bayer, Lumira Ventures, John Mumm, Echo Life Sciences, Viva BioInnovator, Samantha Connor, Plains VC, Alexandria Venture Investments, and AGP Ventures are investors in Deka. J.B.M. is an inventor on the following patents: “Dual cytokine fusion proteins comprising IL-10” (US11292822B2), “Dual cytokine fusions proteins comprising IL-10” (US11572397B2), “Methods for treating malignant tumors with IL-10 variant conjugates” (US10975133B2), and “Antibody variable domain regions fused to IL-10 variant molecules” (US10858412B2).

STAR★Methods

Key resources table

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
CD3 Monoclonal Antibody (OKT3), Functional Grade	Invitrogen	Cat# 16-0037-85; RRID:AB_468855
Ultra-LEAF™ Purified anti-human HLA-A,B,C Antibody	BioLegend	Cat# 311441; RRID:AB_2800815
Human IFN-gamma R1/CD119 Antibody	R&D Systems	Cat# MAB6731; RRID:AB_2122755
Ultra-LEAF™ Purified anti-human CD210 (IL-10 R) Antibody	BioLegend	Cat# 308818; RRID_AB_2800805
eBioscience™ Foxp3/Transcription Factor Staining Buffer Set	Invitrogen	Cat# 00-5523-00
FcR blocking reagent, human	Miltenyi Biotec	Cat# 130-059-901
FcR blocking reagent, mouse	Miltenyi Biotec	Cat# 130-092-575
Ghost Dye™ UV 450 Fixable Viability Dye	Cell Signaling Technologies	Cat# 80862S
Brilliant Violet 570™ anti-human CD4 Antibody	BioLegend	Cat# 300534; RRID:AB_2563791
Brilliant Violet 480 anti-human CD45RA Antibody	BD Biosciences	Cat# 566114; RRID:AB_2739516
BD OptiBuild™ BUV805 Mouse Anti-Human CD45RO	BD Biosciences	Cat# 748367; RRID:AB_2872786
BD OptiBuild™ BUV661 Mouse Anti-Human CD25	BD Biosciences	Cat# 741685; RRID:AB_2871068
BD OptiBuild™ BUV615 Mouse Anti-Human CD95	BD Biosciences	Cat# 752346; RRID:AB_2875863
BD™ CD8 APC-H7	BD Biosciences	Cat# 641409; RRID:AB_1645737
Spark NIR™ 685 anti-human CD62L Antibody	BioLegend	Cat# 304862; RRID:AB_2860809
PE/Cyanine7 anti-human Perforin Antibody	BioLegend	Cat# 308126; RRID:AB_2572049
PE/Fire™ 700 anti-human CD3 Antibody	BioLegend	Cat# 344864; RRID:AB_2876653
BD Horizon™ PE-CF594 Mouse Anti-Ki-67	BD Biosciences	Cat# 567120; RRID:AB_2916453
BD Pharmagen PE Mouse anti-human Granzyme B	BD Biosciences	Cat# 561142; RRID:AB_10561690
BD Horizon™ BB700 Mouse Anti-Human CD27	BD Biosciences	Cat# 566450 AB_2739731
Alexa Fluor® 700 anti-human CD127 (IL-7Rα) Antibody	BioLegend	Cat# 351344 AB_2566200
BD Horizon™ RB780 Mouse Anti-Human FoxP3	BD Biosciences	Cat# 568682; RRID:AB_3662992
BD Horizon™ RB780 Mouse IgG1, κ Isotype Control	BD Biosciences	Cat# 568532; AB_3668760
Brilliant Violet 510™ anti-human CD19 Antibody	BioLegend	Cat# 302242; RRID:AB_2561668
BD Pharmingen™ PE Mouse Anti-Human IRF1	BD Biosciences	Cat# 566322; RRID:AB_2739684
BD Pharmingen™ PE Mouse Anti-Human IRF8	BD Biosciences	Cat# 566373; RRID:AB_2739716
BD Horizon™ Brilliant Stain Buffer Plus	BD Biosciences	Cat# 566385
Zombie Aqua™ Fixable Viability Kit	BioLegend	Cat# 423102
Rat IgG Isotype Control	BioXCell	Cat# BE0088; RRID:AB_3666081
PE/Fire™ 700 anti-mouse CD8a Antibody	BioLegend	Cat# 100791; RRID:AB_2876397
Pacific Blue™ anti-mouse CD25 Antibody	BioLegend	Cat# 102022; RRID:AB_493643
Brilliant Violet 570™ anti-mouse CD62L Antibody	BioLegend	Cat# 104433; AB_10900262
Brilliant Violet 421™ anti-mouse CD11c Antibody	BioLegend	Cat# 117330; RRID:AB_2563099
APC/Fire™ 810 anti-mouse F4/80 Antibody	BioLegend	Cat# 123166; RRID:AB_2894417
Brilliant Violet 605™ anti-mouse Ki-67 Antibody	BioLegend	Cat# 652413; RRID:AB_2562664
PerCP/Fire™ 806 anti-mouse/human CD44 Antibody	BioLegend	Cat# 103082; RRID:AB_3083253
Brilliant Violet 650™ anti-mouse CD19 Antibody	BioLegend	Cat# 115541; RRID:AB_11204087
PE/Fire™ 700 anti-mouse CD4 Antibody	BioLegend	Cat# 100484; RRID:AB_2876395
Alexa Fluor® 700 anti-mouse TCR β chain Antibody	BioLegend	Cat# 109224; AB_1027648
Alexa Fluor® 700 anti-mouse CD3 Antibody	BioLegend	Cat# 100216; AB_493697
APC/Fire™ 750 anti-human/mouse Granzyme B Recombinant Antibody	BioLegend	Cat# 372210; AB_2728377
APC anti-mouse Perforin Antibody	BioLegend	Cat# 154304; AB_2721463
BD Horizon™ PE-CF594 Rat Anti-Mouse Ly-6G and Ly-6C	BD Biosciences	Cat# 562710; AB_2737737
Brilliant Violet 785™ anti-mouse CD8a Antibody	BioLegend	Cat# 100749; RRID:AB_2562610
Brilliant Violet 711™ anti-mouse I-A/I-E Antibody	BioLegend	Cat# 107643; RRID:AB_2565976
FITC anti-mouse CD49b (pan-NK cells) Antibody	BioLegend	Cat# 108906; RRID:AB_313413
BD Pharmingen™ PE Rat Anti-Mouse Siglec-F	BD Biosciences	Cat# 552126; RRID:AB_394341
PE/Fire™ 640 anti-mouse/human CD11b Antibody	BioLegend	Cat# 101280; RRID:AB_2888802
BD Horizon™ BV650 Rat Anti-Mouse CD19	BD Biosciences	Cat# 563235; RRID:AB_2738085
Brilliant Violet 711™ anti-mouse CD4 Antibody	BioLegend	Cat# 100447; RRID:AB_2564586
Anti-mPD-1-mIgG1e3 InvivoFit™	InvivoGen	Cat# mpd1-mab15-1
FcR Blocking Reagent, Mouse	Miltenyi Biotec	Cat# 130-092-575
InVivoMAb anti-mouse CD20	Bioxcell	Cat# BE0356; RRID:AB_2894775
InVivoMAb anti-mouse IL-10R (CD210)	Bioxcell	Cat# BE0050; RRID:AB_1107611
InVivoMAb anti-mouse TNFα	BioXCell	Cat# BE0058; RRID:AB_1107764
InVivoPlus anti-mouse NK1.1	BioXCell	Cat# BP0036; RRID:AB_1107737
InVivoMAb anti-mouse CD8α	BioXCell	Cat# BE0004-1; RRID:AB_1107671
InVivoMAb anti-mouse CD4	BioXCell	Cat# BE0003-1; RRID:AB_1107636
Human IFNγ Antibody Set	Meso Scale Discovery	Cat# B21TT-2
Human TNFα Antibody Set	Meso Scale Discovery	Cat# B21UC-2
Human IL-1β Antibody Set	Meso Scale Discovery	Cat# B21TU-2
Human IL-6 Antibody set	Meso Scale Discovery	Cat# B21TX-2
Human Perforin antibody set	Meso Scale Discovery	Cat# B21AHG-2
Human Granzyme B antibody set	Meso Scale Discovery	Cat# B21APD-2
Human IL-5 antibody set	Meso Scale Discovery	Cat# B21UO-2
Mouse IFNγ antibody set	Meso Scale Discovery	Cat# B22TT-2
Mouse IL-1β antibody set	Meso Scale Discovery	Cat# B22TU-2
Mouse IL-2 antibody set	Meso Scale Discovery	Cat# B22TV-2
Mouse IL-5 antibody set	Meso Scale Discovery	Cat# B22UO-2
Mouse IL-6 antibody set	Meso Scale Discovery	Cat# B22TX-2
Mouse TNFα antibody set	Meso Scale Discovery	Cat# B22UC-2
Mouse Granzyme B antibody set	Meso Scale Discovery	Cat# F223X-3
Biological samples
Human Leukopacks	BioIVT	https://bioivt.com/human-leukopak
Chemicals, peptides, and recombinant proteins
DK210 (EGFR)	Deka Biosciences (This paper)	N/A
DK210 (EBO)	Deka Biosciences (This paper)	N/A
IL-2 (EGFR)	Deka Biosciences (This paper)	N/A
IL-2 (EBO)	Deka Biosciences (This paper)	N/A
IL-10 (EGFR)	Deka Biosciences (This paper)	N/A
IL-10 (EBO)	Deka Biosciences (This paper)	N/A
Recombinant Human IFN-gamma Protein	R&D Systems	Cat# 285-IF/CF
Recombinant Human IL-2 Protein	R&D Systems	Cat# 202-IL/CF
Recombinant Human IL-10 Protein	R&D Systems	Cat# 217-IL/CF
Ficoll-Paque Plus	Cytiva	Cat# 11744003
Geneticin	Gibco	Cat# 10131-035
Lipofectamine™ RNAiMAX	Thermo Fisher Scientific	Cat# 13778075
Purified anti-human CD210 (IL10RA)	Biolegend	Cat# 308818
Human IFN-gamma R1/CD119 Antibody	R&D Systems	Cat# MAB6731
Recombinant human TGFβ	R&D systems	Cat# 302-B2-002/CF
Human IL-10 Protein	KACTUS	Cat# IL1-HM010
Human IL-2 Protein	Prospec	Cat# CYT-209
IFN-gamma Protein, Human	MedChemExpress	Cat# HY-P7025
Lipopolysaccharides from Escherichia coli O111:B4	MilliporeSigma	Cat# L4391
Recombinant Human IL-10 Protein, CF	Biotechne	Cat# 1064-ILB
EBV-IL-10	Biotechne	Cat# 915-VL-010/CF
Fenestra HDVC	MediLumine	SKU: HDVC-121
SuperScript™ IV VILO™ Master Mix	Thermo Fisher Scientific	Cat# 11766050
FTY720	Sigma-Aldrich	Cat# SML0700
Biotinylated Human EGFRvIII Protein, His, Avitag™	ACROBiosystems	Cat# EGFR-H82E0
Biotinylated Human IL-2 R alpha/CD25 Protein, Fc,Avitag™ (MALS verified)	ACROBiosystems	Cat# ILA-H82F9
Biotinylated Human IL-10 R alpha/CD210 Protein, Fc,Avitag™ (MALS verified)	ACROBiosystems	Cat# ILR-H82F6
Evans Blue	Sigma-Aldrich	Cat# E2129
DNase I	Roche	Cat# 11284932001
Puromycin	InvivoGen	Cat# 58-58-2
MSD GOLD Read Buffer B	Meso Scale Discovery	Cat#R60a.m.-3
Critical commercial assays
RNeasy Mini Kit	Qiagen	Cat# 74106
BD Cytofix/Cytoperm™ Fixation/Permeabilization Kit	BD Biosciences	Cat# 554714
Tumor Dissociation Kit, mouse	Miltenyi Biotec	Cat# 130-096-730
Calibrator 1	Meso Scale Discovery	Cat# C0060-2
Calibrator 20	Meso Scale Discovery	Cat# C0328-2
Calibrator 22	Meso Scale Discovery	Cat# C0330-2
Calibrator 23	Meso Scale Discovery	Cat# C0331-2
Calibrator 28	Meso Scale Discovery	Cat# C0411-2
Calibrator 3	Meso Scale Discovery	Cat# C0062-2
Calibrator 9	Meso Scale Discovery	Cat# C0090-2
Calibrator 5	Meso Scale Discovery	Cat# C0065-2
Mouse Granzyme B Calibrator	Meso Scale Discovery	Cat# C023X-2
Diluent 58	Meso Scale Discovery	Cat# R50CA-1
Diluent 3	Meso Scale Discovery	Cat# R50AP-1
Diluent 41	Meso Scale Discovery	Cat# R50AH-1
Diluent 45	Meso Scale Discovery	Cat# R50AI-1
ELISpot Plus: Mouse IFN-γ (HRP)	Mabtech	Cat# 3321-4HPT-2
Seahorse XF T cell Metabolic Profiling Kit	Agilent	Cat# 103772-100
Agilent Seahorse XFe96 Extracellular Flux Assay Kit	Agilent	Cat# 103792-100
Protein Thermal Shift™ Dye Kit	Applied Biosystems	Cat# 4461146
RNeasy Mini Kit	Qiagen	Cat# 74106
Tumor Dissociation Kit, mouse	Miltenyi Biotec	Cat# 130-096-730
TaqMan™ Array, Mouse Immune, Fast 96-well	Thermo Fisher Scientific	Cat# 4418724
TaqMan™ Fast Advanced Master Mix for qPCR	Thermo Fisher Scientific	Cat# 4444557
RNeasy Plus Mini Kit	Qiagen	Cat# 74136
QIAshredder	Qiagen	Cat# 79656
Human IFN-gamma ELISA - Quantikine QuicKit	Biotechne	Cat# QK285
U-PLEX Development Pack, 10-Assay	Meso Scale Discovery	Cat# K15235N-1
Deposited data
Nanostring nCounter data	This paper	Mendeley Data: https://doi.org/10.17632/7skswgr2h5.1https://data.mendeley.com/datasets/7skswgr2h5/1
Experimental models: Cell lines
Green Fluorescent protein expressing SK-BR-3 Cell Line	Innoprot	Cat# P20129
Green Fluorescent protein expressing A-498 Cell Line	Angio-Proteomic	Cat# cAP-0075GFP
B16-F10	ATCC	CRL-6475
Lewis Lung (LL2)	LakePharma	ID: 011000000579068
Experimental models: Organisms/strains
Mouse: C57BL/6J	The Jackson Laboratory	Cat#000664, RRID:IMSR_JAX:000664
Mouse: IFNyKO	The Jackson Laboratory	Cat#002287; RRID:IMSR_JAX:002287
Mouse: BLAB/cJ	The Jackson Laboratory	Cat#000651; RRID:IMSR_JAX:000651
Oligonucleotides
Silencer® Select HLA-A	Thermo Fisher Scientific	Cat# 4392420; Assay ID: s57013
Silencer® Select HLA-B	Thermo Fisher Scientific	Cat# 4392420; Assay ID: s6573
Silencer® Select HLA-C	Thermo Fisher Scientific	Cat# 4392420; Assay ID: s6574
Silencer® Select HLA-B/HLA-C	Thermo Fisher Scientific	Cat# 4392420; Assay ID: s6571
Software and algorithms
Octet BLI Discovery 13.0 Software	SartoriuS	https://www.sartorius.com/download/552418/octet-data-analysis-ht-software-datasheet-en-sartorius-data.pdf
Octet Analysis Studio 13.0 Software	SartoriuS	https://www.sartorius.com/download/552418/octet-data-analysis-ht-software-datasheet-en-sartorius-data.pdf
Mabtech Apex 1.1	Mabtech	https://www.mabtech.com/explore-our-readers
Agilent Seahorse Analytics	Agilent	https://www.agilent.com/en/product/cell-analysis/real-time-cell-metabolic-analysis/xf-software/agilent-seahorse-analytics-787485
IncuCyte® Live-Cell Imaging & Analysis Software	Sartorius	https://www.sartorius.com/en/products/live-cell-imaging-analysis/live-cell-analysis-software
QuantStidioTM Real-Time PCR Software V1.7.2	Applied Biosystems	RRID:SCR_020245
Protein Thermal Shift™ Software & Reagents| Thermo Fisher Scientific – US	Applied Biosystems	https://www.thermofisher.com/us/en/home/life-science/pcr/real-time-pcr/real-time-pcr-applications/real-time-pcr-protein-analysis/protein-thermal-shift.html
Sedfit v.14.1.	Sedfit	https://sedfitsedphat.nibib.nih.gov/software/
SoftMax Pro 6.5	Molecular Devices	https://www.moleculardevices.com/en/assets/tutorials-videos/br/softmax-pro-6-5-acquisition-view
SOFIE - Model GNEX PET/CT - PET- CT Generation Scanner Software	Medical-Xprt	https://www.medical-xprt.com/products/sofie-model-gnext-pet-ct-pet-ct-generation-scanner-796575
AMIDE: Amide’s a Medical Imaging Data Examiner Software	AMIDE	https://amide.sourceforge.net/features.html
MSD Discovery Workbench v4	Meso Scale Discovery	https://software.mesoscale.com/solo/products/ProductOption.aspx?ProdOptionID=1042
nSolver Analysis Software 4.0	NanoString Technologies Inc.	https://nanostring.com/products/ncounter-analysis-system/nsolver-advanced-analysis-software/
BD FACSDiva®	BD Biosciences	https://www.bdbiosciences.com/en-us/products/software/instrument-software/bd-facsdiva-software
SpectroFlow® version 3.1.0	Cytek Biosciences	https://cytekbio.com/pages/spectro-flo
FlowJo_v10.8.1 Software	FlowJo, LLC	https://flowjo.com/docs/flowjo10/getting-acquainted/10-8-release-notes/10-8-1-release-notes
Other
Flat Bottom 96-Well Plates	Caplugs Evergreen	290-8195-Z1F
CD8 MicroBeads, human	Miltenyi Biotec	Cat# 130-045-201
CD56 MicroBeads, human	Miltenyi Biotec	Cat# 130-050-401
CD8 (TIL) MicroBeads, mouse	Miltenyi Biotec	Cat# 130-116-478
AIM V™ Medium	Gibco	Cat# 12055083
Human Serum	Sigma-Aldrich	Cat# H4522-100ML
RPMI-1640 Medium with L-Glutamine	Quality Biological	Cat# 112-025-101
DMEM, high glucose	Gibco	Cat# 11965092
Fetal Bovine Serum	GeminiBio	Cat# 100-500
Costar™ 96-Well, Cell Culture-Treated, Flat-Bottom Microplate	Corning	Cat# 3596
Costar® 24-well Clear TC-treated Multiple Well Plates	Corning	Cat# 3524
Chirascan Q100	AppliedPhotophysics	https://www.photophysics.com/product-pages/chirascan-q100/
Optima XL-I	Beckman Coulter	https://www.beckman.com/landing/ppc/cent/ultra?utm_source=google&utm_medium=cpc&utm_term=analytical%20ultracentrifugation&gclid=EAIaIQobChMIoMWskJ_riQMVRUtHAR2FiR0iEAAYASABEgKXzvD_BwE
CD8 (TIL) MicroBeads, mouse	Miltenyi Biotec	Cat# 130-116-478
Lympholyte®-M Cell Separation Media	Cedarlane	Cat# CL5031
Heparin sodium salt from porcine intestinal mucosa	MilliporeSigma	Cat# H3149
ACK Lysing Buffer	Quality Biological	Cat# 118-156-101
RPMI-1640 Medium	ATCC	Cat# 30-2001
BD Pharmingen™ Stain Buffer (BSA)	BD Biosciences	Cat# 554657
CD8 (TIL) MicroBeads, mouse	Miltenyi Biotec	Cat# 130-116-478
Lympholyte®-M Cell Separation Media	Cedarlane	Cat# CL5031
Corning® Matrigel® Growth Factor Reduced (GFR) Basement Membrane Matrix, LDEV-free	Corning®	Product #354230
Dulbecco’s Modified Eagle’s Medium (DMEM)	ATCC	Cat# 30-2002
RPMI-1640 Medium	ATCC	Cat# 30-2001
Ambion Nuclease-Free Water	Invitrogen	Cat# AM9937
MESO QuickPlex SQ 120MM (with Methodical Mind software)	Meso Scale Discovery	https://www.mesoscale.com/en/products_and_services/instrumentation/quickplex_sq_120mm
NanoDrop™ One Microvolume UV-Vis Spectrophotometer	Thermo Fisher Scientific	Cat#ND-ONE-W
QuantStudio™ 7 Flex Real-Time PCR System, 96-well, desktop	Thermo Fisher Scientific	Cat# 4485690
Albumin from mouse serum	Sigma-Aldrich	Cat# A3139
Glycine 98.5–101.5% FCC	J.T.Baker	Cat# 0581-05
Pluronic® F-68	Gibco	Cat# 24040-032
Sucrose	Fisher Scientific	Cat# S6
Trisodium citrate dihydrate	Sigma-Aldrich	Cat# S1804

Experimental model and study participant details

Cell lines

All mammalian cells were cultured at 37°C in a humidified atmosphere with 5% CO_2. The concentration and viability of the cell lines were evaluated using a Cytek Guava Muse Cell Analyzer (Cytek) or the LUNA-II Automated Cell Counter (Logos Biosystems). Parental cell lines from ATCC, LL/2 (CRL-1642) and B16-F10 (CRL-6457), were transfected at LakePharma to generate cell lines expressing the extracellular and transmembrane domains of human EGFR, B16F10^hEGFR+ and LL2^hEGFR+. These cells were cultured in Dulbecco’s modified Eagle’s medium (DMEM; ATCC or Quality Biological) with 10% fetal bovine serum (FBS; GeminiBio) and 1 μg/mL Puromycin (InvivoGen). The female breast cancer cell line, SK-BR-3 expressing Green fluorescent Protein (SKBR3-GFP) were purchased from (Innoprot, P20129) and cultured in Dulbecco’s modified Eagle’s medium (DMEM; Gibco) with 10% fetal bovine serum (FBS; GeminiBio) and 200 μg/mL Geneticin (Gibco). The human female kidney carcinoma cell line expressing GFP, A-498-GFP (Angio-Proteomic, cAP-0075GFP), were maintained in EMEM (ATCC) with 10% FBS (GeminiBio). Cell lines were purchased at the beginning of these studies from the specified vendors but were not further verified or tested for mycoplasma. Human primary cells were purified from healthy donor Leukopaks purchased from BioIVT and collected under IRB-approved protocols. Primary cells were collected from both male and female individuals then divided for treatment with each condition that is depicted for each study.

Animals

Mouse studies were performed as approved by the Institutional Animal Care and Use Committee (IACUC) of Deka Biosciences, that of the University of California, Los Angeles, or that of Noble Life Sciences in accordance with where the study was conducted and complied with the National Research Council’s Guide for the Care and Use of Laboratory Animals. For studies conducted at Deka Biosciences wild-type C57BL/6 (RRID:IMSR_JAX:000664), B6.129S7-Ifng^tm1Ts/J (IFNγ−/−) (RRID:IMSR_JAX:002287), and B6.129S-Tnf^tm1Gkl/J (TNFα−/−) (RRID:IMSR_JAX:005540) female 8–12-week-old mice were purchased from The Jackson Laboratory and maintained under climate controlled, specific pathogen-free conditions in the animal facilities of Deka Biosciences. All animals were randomly assigned to experimental groups. A 1-Month Toxicity, Toxicokinetic and Pharmacodynamic Study (Study No. 20330957) using both male and female approximately 2-year-old purpose-bred, naive Cynomolgus monkeys (Macaca fascicularis) was conducted at Charles River Laboratories (CRL) following Good Laboratory Practices (GLP). The Study was performed in accordance with the U.S. Department of Health and Human Services, Food and Drug Administration (FDA), United States Code of Federal Regulations, Title 21, Part 58: Good Laboratory Practice (GLP) for Nonclinical Laboratory Studies and as accepted by Regulatory Authorities throughout the European Union (OECD Principles of Good Laboratory Practice), Japan (MHLW), and other countries that are signatories to the OECD Mutual Acceptance of Data Agreement. The study was performed in accordance with Charles River - Nevada Institutional Animal Care and Use Committee (IACUC) and complied with the National Research Council’s Guide for the Care and Use of Laboratory Animals.

Method details

Protein production

All engineered proteins for these studies that are not commercially available were produced at either Deka Biosciences, Inc. or Cytovance Biologics, Inc. The proteins were expressed in stably transfected Chinese hamster ovary cells, and purified using a multimodal cation exchange capture chromatography followed by strong anion and cation exchange polishing chromatographies. Concentration and buffer exchange into formulation buffer was performed using flat-sheet ultrafiltration & diafiltration. Strength, potency, purity, and identity of the study materials were assessed using absorbance at 280nm, IL-2/IL-10 bioassay, EGFR-binding biolayer interferometry, size-exclusion chromatography, capillary sodium dodecyl sulfate (SDS) electrophoresis, and capillary isoelectric focusing, respectively. All analytical results passed pre-determined acceptance criteria.

Cytokines and reagents

Fusion proteins conjugated to ScFv domains including both the untargeted molecules (i.e., IL-2 (ScFv), IL-10 (ScFv), DK2¹⁰ (ScFv), and EBV-IL-10 (ScFv) proteins) and those with the anti-EGFR CDR sequences from cetuximab (namely, IL-2 (EGFR) and IL-10 (EGFR), IL-2 (EGFR)) were generated by Deka Biosciences. The EBV IL-10 was from Biotechne. In human primary cell studies, recombinant human cytokines including IL-2 and IL-10 purchased from R&D Systems were used, and recombinant human IFNγ from MedChemExpress. In murine studies performed at Deka Biosciences, recombinant human IL-2 (Prospec) and recombinant human IL-10 (KACTUS) were used. High affinity IL-10 muteins sequences were derived from a previous publication where the Kd values are reported.²⁹ Our EBV IL-10 Mutein 2 and the DK2¹⁰ (EGFR) contains the two mutations corresponding to V43L/A87I mutein of this publication.

In vivo tumor models

B16F10^hEGFR+ cells (4.5 x 10⁵) and LL2^hEGFR+ cells (6.5 x 10⁵) were subcutaneously implanted in female C57BL/6 mice of 8–12 weeks of age (The Jackson Laboratory). Tumors were allowed to become established (approximately 100mm³ for B16F10^hEGFR+ and 50mm³ for LL2^hEGFR+) before depleting B cells by intravenous or intraperitoneal administration of 200-250μg anti-CD20 (Biolegend). After 24 h, animals were randomized and treatment was initiated. After initiation of treatment, tumor measurements were taken 3 times per week using calipers. Tumor volume was calculated using the following formula: V = 0.5 x Length x Width.²

In tumor efficacy studies, B16F10^hEGFR+ tumor-bearing mice were subcutaneously dosed with either human recombinant, scFv, or EGFR-targeted molecules. All scFv and EGFR-targeted molecules were formulated in vehicle control buffer (10.2 mM sodium citrate, 133 mM glycine, 5% sucrose, 0.01% Pluronic F-68, and 0.05% mouse serum albumin, pH 7.3). IL-2 and IL-10 were administered, independently or in combination, at varying doses to assess therapeutic efficacy. Dosing continued until DK2¹⁰ (EGFR)-treated tumors approached 0 mm³. Immunophenotyping of immune cell subsets was conducted two hours after the final administration.

Cell type ablation and tumor rechallenge studies were performed at Noble Life Sciences. Animals were administered once weekly via intraperitoneal injection with 0.35 mg anti-CD8 (clone 53-6.7), 0.25 mg anti-CD4 (clone GK1.5), 0.3 mg anti-asialo (PK136), or 0.35 mg/mouse isotype control (Rat IgG2a) (all BioXCell). For FTY720 studies, animals were given 40 μg DK2¹⁰ (EGFR), 20 μg FTY720 (Sigma-Aldrich) reconstituted in DMSO, the combination of DK2¹⁰ (EGFR) and FTY720, or an equivalent volume of vehicle control 3 times per week. DK2¹⁰ (EGFR) and vehicle control were administered subcutaneously and FTY720 was administered intraperitoneally. To evaluate the dependency of DK2¹⁰ (EGFR) on endogenous IFNγ, B16F10^hEGFR+ tumor-bearing IFNγ knockout or wild-type C57BL/6J mice (The Jackson Laboratory) were subcutaneously administered with 40 μg DK2¹⁰ (EGFR) or equivalent volume of vehicle control 3 times per week. In primary tumor outgrowth study, tumor outgrowth was assessed for ∼1 month in animals exhibiting “cured” tumors after which all animals were rechallenged with the parental tumor.

Lewis lung murine melanoma studies were conducted at Noble Life Sciences. Female 8–12-week-old C57BL/6J mice (The Jackson Laboratory) were given intraperitoneal administrations of 2.5 μg anti-PD-1 (InvivoGen) once weekly alone or in combination with 2 mg/kg DK2¹⁰ (EGFR) dosed subcutaneously 3 times per week for 1 week. Subsequent tumor outgrowth was evaluated.

The patient-derived xenograft study was conducted at The Jackson Laboratory. A patient derived medium EGFR-expressing non-small cell lung cancer xenograft tumor model was established in human CD34⁺ cord-blood engrafted NOD scid gamma animals (NSG; The Jackson Laboratory). Once the tumors reached ∼150 mm³, anti-PD-1 (Pembrolizumab) was administered intraperitoneally at 10 mg/kg once every 5 days, or DK2¹⁰ (EGFR) at 0.5 or 2 mg/kg was dosed subcutaneously 3 times per week.

In vivo IL-2 induced toxicity assays

For serum cytokine analysis, two-to three-month-old female C57BL/6 mice (The Jackson Laboratory) were randomized into five groups. Mice were intraperitoneally administered 250 μg anti-CD20 (Biolegend). One day later, each group began treatment by receiving a 100 μL subcutaneous dose of either vehicle control, 5 μg rhIL-2 (Prospec), 2.4 μg rhIL-10 (Sino Biological), or a combination of 5 μg rhIL-2 and 2.4 μg rhIL-10. Mice were administered a dose every 24 h, and the study lasted 9 days. Serum was collected after 1, 3, 6, or 9 doses, approximately four hours from the most recent dose on that day.

To assess pulmonary edema, female wild-type C57BL/6J, B6.129S7-Ifng^tm1Ts/J (IFNγ−/−), and B6; 129S-Tnf^tm1Gkl/J (TNFα−/−) (The Jackson Laboratory) mice of 10–12 weeks were intraperitoneally administered with 100 μg rhIL-2 (Prospec), a combination of 100 μg rhIL-2 and 200 μg rhIL-10 (KACTUS), 40 μg DK2¹⁰ (EGFR), or an equivalent volume of vehicle control twice a day for a total of 7 doses. For TNFα neutralization studies, animals were given intraperitoneal administrations of anti-mouse TNFα (BioXCell) or Rat IgG1 isotype control (BioXCell) 1 day prior to the first administration of rhIL-2 (day −1) and on days 1 and 3. Two hours after the final administration of cytokines, 0.5% Evans Blue dye (Sigma-Aldrich) was intravenously administered via the tail vein. Thirty minutes following Evans Blue dye administration, transcardiac perfusion was performed with 50 mL of PBS containing 10 units/ml heparin (Sigma-Aldrich) at a rate of 10 mL/min. Lungs were harvested, weighed, and placed in 2 mL of formamide for 24 h at 37°C. The supernatant was collected, and Evans Blue concentration was analyzed by spectrophotometry (620 nm) using a Spectra Max M2.

Isolation of tumor infiltrating lymphocytes

To isolate TILs, tumor tissue was processed by mechanical and enzymatic dissociation according to the Mouse Tumor Dissociation Kit (Miltenyi Biotec) protocol. Briefly, tumor tissues were manually chopped in RPMI (ATCC) and transferred to a gentle MACS C-Tube with the provided enzymes and 0.2 μg/mL DNase I (Millipore Sigma). The samples were digested using the 37C_m_TDK_1 program on a gentleMACS Dissociator then placed on ice. Single-cell suspensions were filtered through a 70 μm strainer with RPMI +0.2 μg/mL DNase I, then a 40 μm strainer. Samples were centrifuged at 300 x g for 8 min and re-suspended in RPMI. Lymphocytes were isolated by density gradient centrifugation using Lympholyte-M (Cedarlane) at 1300 x g for 20min at 20°C then washed in RPMI.

To subsequently isolate CD8⁺ cells for ELISpot and Seahorse assays, single-cell suspensions of TIL were washed and re-suspended in PBS supplemented with 2.5% bovine serum albumin (BSA). Cells were labeled with CD8 (TIL) MicroBeads (Miltenyi) for 15 min at 4°C then placed in LS columns for MACS Separator positive isolation. Cells were collected and re-suspended in RPMI for subsequent studies. For nCounter analysis, tumor-infiltrating lymphocytes were isolated as described above and stained for viability (Zombie Aqua) and surface markers to sort effector memory CD8⁺ T cells (CD45⁺/CD3⁺/CD4^-/CD8⁺/CD44⁺/CD62L⁻) using a BD FACSAria III.

nCounter analysis of tumor infiltrating effector memory CD8⁺ T cells

Cell lysates were prepared by re-suspending sorted effector memory CD8⁺ TILs in 350 μL Buffer RLT provided in the RNeasy Mini Kit (Qiagen), vortexing for 1 min, and freezing at −80°C prior to RNA extraction and nCounter analysis. Total RNA from sorted cells was extracted using the Qiagen RNeasy Mini Kit according to manufacturer recommendations. The purity and concentration of RNA was measured using a NanoDrop One (Thermo Scientific). Gene expression analysis using a nCounter Pro Analysis System was performed by Bruker (Seattle, WA) following the manufacturer’s instructions. RNA from each sample was hybridized using the PanCancer IO 360 Panel of 770 genes. Absolute read counts were extracted using the nSolver Analysis Software (version 4.0). Target genes were normalized to 12 housekeeping genes (Abcf1, Dnajc14, Ercc3, G6pdx, Gusb, Oaz1, Polr2a, Psmc4, Pum1, Sdha, Tbc1d10b, Tmub2).

Quantitative PCR

Quantitative PCR was performed according to the manufacturer’s protocol after RNA extraction and cDNA synthesis. RNA was extracted from tumors using Qiagen QIAshredder tubes to homogenize tumor tissue lysates and Qiagen RNeasy Mini Kit to purify RNA in the TME. RNA was converted to cDNA using Invitrogen SuperScript IV VILO Master Mix with ezDNase enzyme. cDNA was analyzed using a QuantStudio 7 Flex with TaqMan Array, Mouse Immune, Fast 96-well plates. cDNA was diluted to 25 ng per reaction well of the TaqMan Array, Mouse Immune plate and added to TaqMan Fast Universal PCR Master Mix (2✕) according to manufacturer’s instructions. Samples were run for 40 cycles (95°C for 0:03, 60°C for 0:30). Data were analyzed using the 2^(−ΔΔcT) method.

Isolation of human primary PBMC

Human primary PBMCs were isolated from healthy donor leukopaks (BioIVT) using Ficoll-Paque (Cytiva) density gradient centrifugation and used fresh or cryopreserved in liquid nitrogen until later use.

PBMC cytokine stimulation

Isolated PBMC were seeded at 2.0 × 10⁵ cells/well in a 96-well, tissue culture (TC)-treated plate (Corning) in AIM-V medium (Gibco) containing 2% human serum (Sigma Aldrich). For studies pertaining to Figure 1, control cells were left untreated, whereas other cells received a dose titration (0–13 nM) of DK2¹⁰ (EGFR), IL-2 (R&D Systems), IL-10 (R&D Systems), or IL-2+IL-10 for 24 h with some receiving further stimulation with 1 μg/mL anti-CD3 antibody (Invitrogen) for an additional 24 h. Cell supernatants were collected prior to and after anti-CD3 stimulation for analysis of secreted factors using the MSD platform.

For receptor blocking experiments, PBMC were plated as described above and left untreated (control) or were treated with either IL-2 (20 ng/mL), IL-2 + anti-human IFNγR1 (5 μg/mL), IL-2 + anti-human IL-10RA (30 μg/mL), or IL-10 (50 ng/mL). Receptor blocking antibodies were added to culture 1 h prior to the addition of IL-2, as indicated. Cell supernatants were collected at the indicated time points up to 48 h.

In vitro assays for response to IFNy

Human PBMCs were plated at 2 x 10⁵ cells/well on a 96-well TC plates in AIM V containing 2% HSA. To analyze IRF1 and IRF8 protein expression in monocytes, monocytes were isolated as previously described.⁷⁵ Briefly, human PBMCs were plated at 5 x 10⁶ cells/well on a 48-well TC plate in AIM V containing 2% HSA followed by several rounds of washing with media to remove non-adherent cells. Where IL-10 (100 ng/mL) was added, this was added to culture 4 h prior to the addition of IFNy (10 ng/mL). From PBMCs, supernatant was collected at incremental time points up to 48 h after addition of cytokines and frozen at −20°C until analysis by MSD. To assess intracellular IRF1 and IRF8 levels, Accutase (STEMCELL Technologies) was used to remove adherent cells followed by the fixation and permeabilization of cells for intracellular staining with either PE labeled anti-IRF1 or anti-IRF8. Protein expression was compared to untreated cells (negative) and IFNγ alone treated cells (control) by determining the median fluorescence intensity (MFI) of monocytes and calculating the relative change in MFI from the negative as compared to the control [100 x (MFI of treated – MFI of negative)/(MFI of control – MFI of negative)]. Statistical difference was performed using a one-sample t-test to determine a difference from the IFNγ-treated controls.

CD8⁺ T cell cytolysis of tumor cells

CD8⁺ T cells were isolated fresh from healthy donor leukopaks (BioIVT) by positive selection using human CD8 microbeads (Miltenyi) according to the manufacturer’s instructions. Isolated CD8⁺ T cells were plated at a density of 2.5 × 10⁶ cells/mL in 24-well plates (Corning) and cultured in AIM-V medium with or without DK2¹⁰ (EGFR) (100 ng/mL), IL-2 (20 ng/mL), IL-10 (50 ng/mL), or the IL-2+IL-10 (20 or 50 ng/mL respectively) for 48 h. Medium with these reagents is referred to as condition medium. Following this pre-treatment, the CD8⁺ T cells were co-cultured with SKBR3-GFP cells (seeded 24 h prior at 2.5–3.0 × 10⁵ cells per well in DMEM condition medium with 10% FBS (GeminiBio) at 10:1 effector-to-target cell ratio (E:T). Cytolysis was monitored over 5 days (1 round) using the IncuCyte S3 Live-Cell Analysis System (Essen Bioscience/Sartorius). From these co-cultures of CD8⁺ T cells and SKBR3-GFP cells, supernatant was collected to analyze soluble factors using multiplex MSD platform, the metabolic activity of CD8⁺ T cells assessed using Seahorse XF metabolic analyzer (Agilent), and CD8⁺ T cell phenotype profiled by flow cytometry. The condition media for these co-cultures were refreshed every 2–3 days. To investigate persistent CD8⁺ T cell cytotoxicity of DK2¹⁰ (EGFR) treated cells, CD8⁺ T cells from the indicated conditions after each 5-day round were harvested, washed, and reseeded at a 10:1 E:T with freshly seeded SKBR3-GFP cells in the same condition medium. A total of 5 rounds of were performed.

For the assessment of MHC-I mediated CD8⁺ T cell cytotoxicity, siRNA oligonucleotides targeting HLA-A (s57013), HLA-B (s6573), HLA-C (s6574), and HLA-B/C (s6571) (100 nM) (Thermo Fisher Scientific) were transfected into SK-BR3-GFP cells using Lipofectamine RNAiMAX (Thermo Fisher Scientific) following the manufacturer’s instructions. Transfections were carried out 48 h prior to co-culture with CD8⁺ T cells taken from the end of the third round of cytolysis from either untreated controls of DK2¹⁰ (EGFR)-treated conditions. Additionally, siRNA-transfected SKBR3-GFP cells were incubated with an HLA-ABC blocking antibody (10 μg/mL) (Invitrogen) in serum-free medium for 45 min prior to the addition of CD8⁺ T cells. Cytotoxicity was monitored over 48 h.

Natural killer (NK) cell-mediated cytotoxicity

NK cells were isolated fresh from healthy donor leukopaks (BioIVT) by positive selection using human CD56 microbeads (Miltenyi) according to the manufacturer’s instructions. Isolated NK cells were plated at a density of 2.5 x 10⁶ cells/mL in 24-well plates (Corning) and cultured for 48 h in AIM-V medium with or without DK2¹⁰ (EGFR) (100 ng/mL), IL-2 (20 ng/mL; R&D Systems), IL-10 (50 ng/mL; R&D Systems), or IL-2+IL-10 (20 and 50 ng/mL respectively) for 48 h. Following this pre-treatment, the NK cells were co-cultured with SKBR3-GFP cells in AIM-V medium at 10:1 effector-to-target cell ratio. For these studies, NK cells were seeded 24 h prior at 3.0 × 10⁵ cells per well in DMEM with 10% FBS (GeminiBio). Cytolysis was monitored over 48 h using the IncuCyte S3 Live-Cell Analysis System (Essen Bioscience/Sartorius).

In vivo biodistribution profiling by microPET/CT imaging

In vivo biodistribution studies were conducted at the University of California, Los Angeles. DFO conjugation and ⁸⁹Zr radiolabeling were performed as described.⁷⁶ Briefly, DFO-NCS was first conjugated to DK2¹⁰ (EGFR) and DK2¹⁰ (scFv) via reaction for 30 min at 37°C, followed by the addition of ⁸⁹Zr⁴⁺ solution and incubation for 60 min at room temperature. The radiochemical purity of the labeled products was verified using radio-TLC.

B16F10^hEGFR+ mouse melanoma cells were subcutaneously injected in the flank of 8-week-old female C57BL/6J mice (The Jackson Laboratory). When tumors reached 300 mm³, animals were anesthetized and intravenously injected with 100 μL CT contrast agent Fenestra HDVC agent (MediLumine), immediately followed by microPET (energy window 350–650 keV, 10-min static scan) and microCT (voltage 80 kVp, current 150 μA, 720 projections, 200μm resolution, scan time 1 min) imaging on a GNEXT PET/CT scanner (Sofie Biosciences, Dulles, VA), which is the 0 h timepoint. Animals were then subcutaneously injected with 50 μCi ⁸⁹Zr-DFO-DK2¹⁰ (EGFR) or ⁸⁹Zr-DFO-DK2¹⁰ (scFv) at 2 mg/kg, followed by microPET/CT imaging at 6, 24, 48, and 72 h post-injection. The PET images were reconstructed using a 3D-Ordered Subset Expectation Maximization (OSEM) algorithm (24 subsets and 3 iterations), with random, attenuation, and decay correction. The CT images were reconstructed using a Modified Feldkamp Algorithm. Amide software⁷⁷ was used to analyze co-registered PET/CT images and an ellipsoid-shape region of interest (ROI) was placed inside the tumor in the PET images for quantifying tumor accumulation.⁷²

Multiplex immunoassay

Quantification of cytokine levels was conducted using custom U-PLEX Multiplex Biomarker Kits (Meso Scale Discovery). All experiments were performed according to the manufacturer’s recommended protocol. Briefly, U-PLEX linkers were individually conjugated to biotinylated capture antibodies. Conjugated linkers were combined and used to coat multi-spot MSD 96-well plates. Then samples or Standards were added and plates were incubated for another two hours. Following this incubation, SULFO-TAGGED detection antibody were added and incubated for one hour. Lastly, MSD Read buffer B was added to all wells and plates were read immediately using a MESO QuickPlex SQ 120MM reader. Following each step, plates were washed 3 times with PBS-T (PBS with 0.05% tween) and each incubation was done with plates shaking (700 rpm) at room temperature. Data were analyzed using MSD Workbench software (v 4.0.12) according to the manufacturer’s protocol.

In vitro cell response assay

CD8⁺ T cells were plated in AIM V media at 2.5 x 10⁶ cells/well in a 24 well TC-plate and exposed for 24 h to a dose titration (0–5.5 nM) of either IL-10, EBV-IL-10 (Biotechne) or scFv muteins (Deka). Secreted IFNγ was quantified by ELISA (Biotechne). Monocytes plated at 2 x 10⁵ cells/well were exposed for 1–1.5 h to a dose titration (0–5.5 nM) of either IL-10, EBV-IL-10 (Biotechne) or scFv muteins (Deka). Following this incubation, cells were then treated with 10 ng/mL LPS for 12–16 h. Secreted TNFα was quantified by ELISA (Biotechne). The reduction of TNFα was determined as a percentage of the concentration relative to the LPS alone treated controls.

In vitro human regulatory T cell assay

Total T lymphocytes were isolated from healthy donor leukopaks (BioIVT) by positive selection using human CD3 microbeads (Miltenyi) according to the manufacturer’s instructions. T cells were plated at 1 x 10⁶ cells/well on a 48-well TC plate in AIM V Medium (Gibco). T cells were activated with plate bound anti-CD3 (2 μg/mL) and soluble anti-CD28 (1 μg/mL). Cells were cultured for 5 days without cytokines or in the presence of rhIL-2, rhIL-10, DK2¹⁰ (EGFR), and/or rhTGFβ. The molar concentrations of rhIL-2 and rhIL-10 were matched to those cytokine components of DK2¹⁰ (EGFR). The final concentrations of cytokines or DK2¹⁰ (EGFR) were as follows: rhIL-2 (1 ng/mL), rhIL-10 (2.5 ng/mL), DK2¹⁰ (EGFR) (5 ng/mL), and rhTGFβ (10 ng/mL). Following 5 days of culture, cells were analyzed by flow cytometry for the frequency of Foxp3⁺ Treg (CD3⁺/CD4⁺/CD8^-/CD25⁺/CD127^low/Foxp3⁺).

Human primary cell and murine study immune phenotyping by flow cytometry

For flow cytometry analysis of murine samples, spleens and tumors from treated mice were collected on indicated days. Spleens were mashed through 40μm Corning cell strainers followed by lysis of red blood cells with ACK buffer (Lonza Bioscience) for 5 min, quenching with RPMI containing 10% FBS, and washed prior to staining. For TIL, tumors were excised, fragmented and digested using the Mouse Tumor Dissociation Kit (Miltenyi Biotec) protocol followed by the isolation of lymphocytes as described earlier and then staining. For human primary cells, staining was done on cells that had been isolated and cultured as described under their respective study method sections.

Isolated cells were washed in FACS Buffer (3% FBS, 1 mM EDTA, HBSS) or BD FACS Buffer prior to staining. First cells were incubated with FcR block, mouse or human (Miltenyi Biotec) for 10 min at 4°C. Viability staining was performed using Ghost Dye UV450 Fixable Viability Dye (Cell Signaling Technologies) for human samples or Zombie Aqua Fixable Viability dye (BioLegend) for murine samples by incubating 30 min at 4°C along with antibodies against surface markers. For human studies T cell subsets were identified using previously described phenotypic markers,³³ anti-human CD4 BV570 (Biolegend, 300534), CD45RA BV480 (BD, 566114) CD45RO BUV805 (BD, 748367), CD25 BUV661 (BD, 741685), CD95 BUV615 (BD, 752346), CD8 APC-H7 (BD, 641409), CD62L Spark-NIR 685 (BioLegend, 304862), Perforin PE-Cy7 (BioLegend, 308126), CD3 PE-Fire 700 (BioLegend, 344864), Ki-67 PE-CF594 (BD, 567120), Granzyme B PE (BD, 561142), CD27 BB700 (BD, 566450), CD127 AF700 (BioLegend, 351344), FoxP3 RB780 (BD, 568682), Mouse IgG1, κ Isotype Control RB780 (BD, 568532), CD19 BV510 (BioLegend, 302242), IRF1 PE (BD, 566322), IRF8 PE (BD, 566373). For staining in murine studies, the following antibodies were used: anti-mouse CD8α PE-Fire 700 (BioLegend, 100791), CD8α BV785 (BioLegend, 100749), CD25 Pacific Blue (BioLegend, 102022), CD62L BV570 (BioLegend, 104433), CD11c BV421 (BioLegend, 117330), F4/80 APC-Fire 810 (BioLegend, 123166), Ki-67 BV605 (BioLegend, 652413), CD44 PerCP/Fire 806 (BioLegend, 103082), CD19 BV650 (BioLegend, 115541), CD19 BV650 (BioLegend, 563235), CD4 PE/Fire 700 (BioLegend, 100484), CD4 BV711 (BioLegend, 100447), TCRβ AF700 (BioLegend, 109224), CD3 AF700 (Biolegend, 100216), Granzyme B APC/Fire 750 (BioLegend, 372210), Perforin AOC (BioLegend, 154304), Ly6G/Ly6C PE-CF594 (BD, 562710), CD49b FITC (BioLgend, 108906), Siglec-F PE (BD, 552126), CD11b PE/Fire 640 (BioLegend, 101280). Following viability and surface staining, cells were washed twice with FACS buffer to be fixed with 4% paraformaldehyde. When intracellular staining was required, cells were fixed and permeabilized with a commercially available permeabilization/fixation buffer, either BD Cytofix/CytoPerm Fixation/Permeabilization Kit (BD, 554714) or eBioscience Foxp3/Transcription Factor Staining Buffer Set (Invitrogen, 00-5523-00) following the manufacturer’s specifications. Intracellular staining was performed for 1 h at 4°C with antibodies diluted in the corresponding kit permeabilization buffer. Samples were then washed twice with FACS Buffer before analysis. All mouse samples were analyzed on a Northern Lights spectral flow cytometer (Cytek). Human samples were analyzed on an Aurora spectral flow cytometer (Cytek). All data were analyzed using FlowJo Software (FlowJo 10.8.1). Gating strategies are depicted in supplementary figures.

Seahorse assay

One day prior to the metabolic assay, the sensor cartridge from the Agilent Seahorse XFe96 Extracellular Flux Assay Kit was hydrated with calibrant solution and incubated overnight at 37°C in a non-CO₂ incubator. On the day of the assay, Seahorse XF RPMI assay medium and compounds from the Agilent Seahorse XF T cell Metabolic Profiling Kit, including Oligomycin A, Bam15, and Rotenone/Antimycin A, were prepared according to the manufacturer’s instructions. CD8⁺ T cells, harvested after a 5-day co-culture with SKBR3-GFP cells, were resuspended in Seahorse XF RPMI assay medium and plated at a density of 2.0 x 10⁵ cells/well on a Seahorse XF poly-D-lysine (PDL) coated cell culture plate. For the assessment of tumor-infiltrating CD8⁺ T cells from B16F10^hEGFR+-tumor bearing mice, CD8⁺ TILs were isolated as described earlier. Isolated cells were seeded at 2.0 x 10⁵ cells/well in 96-well tissue culture-treated plates in RPMI media (Quality Biologics) supplemented with 10% FBS and incubated overnight to rest them. The following morning, the medium was replaced with serum-free RPMI for 4 h. After resting, the cells were counted and plated at 1.0 x 10⁵ cells/well in Seahorse XF RPMI medium on the Seahorse XF PDL-coated plate. Upon completion of the assays, data were analyzed using the Seahorse Analytics software (Agilent).

ELISpot

ELISpot assays from MabTech were performed according to manufacturer recommendations. Briefly, tumor-infiltrating CD8⁺ T isolated from tumor-bearing mice as previously described were rested for 4–12 h in RPMI +10% FBS. After resting, the CD8⁺ cells were plated on an ELISpot plate with wells pre-coated with IFN-γ capture antibody (mAb AN18). Cognate tumor cells were exposed to 10 ng/mL IFN-γ for 12 h prior to co-culture with isolated CD8⁺ cells at an effector to target ratio of 10:1 in the ELISpot plate. Background signal was determined from conditions in which CD8⁺ cells were plated without tumor cells. After plating, cells were cultured for 48 h in RPMI +5% FBS at 37^oC, 5% CO2 before developing plates according to manufacturer recommendations. Briefly, cells were washed with PBS five times prior to the addition of 1 μg/mL R4-6A2-biotin capture antibody in PBS with 2% serum. After incubating for two hours, wells were washed five times in PBS. Next 1 μg/mL streptavidin-HRP was added and incubated for one hour. Lastly, plates were again washed five times, and spots were developed with provided TMB substrate. Spots were quantified using a MabTech IRIS ELISpot reader with Mabtech Apex 1.1 software.

Binding affinity measurement by octet BLI

The binding affinities of DK2¹⁰ (EGFR) and its various domains against their cognate receptors were measured by Sartorius Octet biolayer interferometry (BLI) reader. The sensors were pre-hydrated in PBST buffer for 10 min before loading into the OCTET system. Immobilization of biotinylated hEGFRvIII, hIL-10RA, or hIL2Rα (Each from AcroBiosystems) onto streptavidin (SA) biosensor tips was achieved by dipping the sensors into wells containing 2 μg/mL of respective receptors in PBST for 300 s, followed by washing in PBST for 60 s to remove unbound ligand. Following immobilization, the biosensor tips were transferred to wells containing PBST to establish a baseline signal. The baseline was monitored for 60 s to ensure signal stability prior to the binding assay. For the binding interaction analysis, ligand proteins were prepared in PBST. The immobilized receptor biosensors were dipped into wells containing the DK2¹⁰ (EGFR), IL-2, or IL-10 at the specified concentrations for 300 s to measure the association phase. After association, the biosensors were moved to wells containing PBST alone to monitor dissociation for 1000 s. Data acquisition and analysis were performed using the Octet BLI Discovery 13.0 and Data Analysis software, Octet BLI Analysis 13.3. Kinetic parameters, including the association rate constant (ka), dissociation rate constant (kd), and equilibrium dissociation constant (K_D), were determined by fitting the binding curves to a 1:1 binding model.

Far- and near-ultraviolet circular dichroism

For far-UV analysis, the samples were thawed, diluted to 1.6 mg/mL, and dialyzed in dialysis buffer (10.2 mM sodium citrate, 5% sucrose, 0.01% Pluronic F-68, pH 7.30) to minimize interference from glycine in the formulation buffer. It was then dialyzed at 2°C–8°C for 20 h using the “Float-A-Lyzer” dialysis device for 0.4 mL volume. For near-UV analysis, the samples were thawed and diluted to 0.8 mg/mL. 0.1- or 10-mm cuvettes were used for far UV or near-UV analysis respectively. BSA was diluted with 1X PBS from 10 to 1.6 mg/mL for far-UV and to 0.8 mg/mL for near-UV. The diluted BSA solutions served as controls preceding (BSA Initial) and following (BSA Final) the analysis. Spectra were collected in triplicate, blank-subtracted, normalized, converted to CD units (Δε) and average was reported. Mean residue ellipticity (MRE) was calculated using MRE = 100 × θ/(CMR × l), where CMR is the mean residue concentration Samples were analyzed from 250 to 350 nm for near-UV and from 180 to 260nm for Far-UV on the Chirascan Q100.

Differential scanning fluorimetry

The thermal stability of DK2¹⁰ (EGFR) was measured via differential scanning fluorimetry (DSF) using the Protein Thermal Shift (Applied Biosystems) assay and software. DK2¹⁰ (EGFR) was diluted to 0.32 mg/mL using drug substance formulation buffer. 12.5 μL of sample, 5 μL of Protein Thermal Shift Buffer, and 2.5 μL of 20x diluted Protein Thermal Shift Dye were combined and mixed in a 96-well PCR plate. The plate was spun at 1000 rpm for 1 min. A melt curve was performed on an Applied Biosystems QuantStudio 7 Flex PCR instrument with a temperature ramp of 0.05°C/s from 25°C to 99°C. Data was analyzed using Protein Thermal Shift Software and melting temperature (Tm) was determined using the first derivative curve.

Sedimentary velocity analytical ultracentrifugation

Sedimentation velocity experiments were conducted with an Optima XL-I (Beckman Coulter) centrifuge using an An60 Ti four-hole rotor. Standard double-sector Epon centerpieces equipped with sapphire windows contained 400 μL of the samples at a 1/10 dilution (17.2 mg/mL) of the solution into the supplied buffer. Interference data were acquired in the continuous mode at a rotor speed of 50,000 rpm at a temperature of 20°C with systematic noise subtracted without averaging and with radial increments of 0.003 cm using the interferometer detector. The density and viscosity of the buffer and the partial specific volume of the protein were calculated using Sednterp. Multi-component sedimentation coefficient distributions were obtained from 100 scans by direct boundary modeling of the Lamm equation using Sedfit (v.14.1). Raw data were analyzed using Sedfit software.

Non-human primate studies

Male, and female, non-naïve, 22 to 60 months (males) and 22 to 48 months (females) of age purpose-bred cynomolgus monkeys were treated with DK2¹⁰ (EGFR) at 0.25, 1.0, and 2.5 mg/kg dosed subcutaneously 3 times a week for 4 weeks. Each dose level consisted of 3–5 male and 3–5 female animals. Animals were housed in accordance with specifications denoted 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 (NCR, Current Edition). The animals were separated during designated procedures/activities or as required for monitoring and/or health purposes as deemed appropriate by Charles River Study director and/or Clinical Veterinarian. The rooms used were documented and kept in study 20330957 records. Pharmacokinetic sampling were taken on the indicated days relative to the first dose of DK2¹⁰(EGFR) and time from the most recent dose also indicated in hours (h). Plasma samples were analyzed using an MSD platform assay that captures IL-10 and detects lL-2. DK2¹⁰ (EGFR) levels are plotted for each dose level. Plasma cytokine concentrations were determined for IFNγ, TNFα, IL-1β, and IL-6 using an MSD U-plex with a lower limit of quantification (LLOQ) of 150 pg/mL and an upper limit of quantification (ULOQ) of 6000 pg/mL. Concentrations were generally below the LLOQ at all time points evaluated or were within the range of pre-dose and/or control. Immunophenotyping of peripheral immune populations was performed at multiple timepoint over the course of the study (Week −1, Day 6 24 h post-dose and 72-h post dose, and Day 28 24 h post-dose and 72 h post-dose) using a CRL validated analytical methods (AP.I-002242.FC4 Panel 1b, AP.20320745.FC1). Cell counts were determined using BD TruCount tubes in combination with CD45 labeling. Briefly, whole blood was collected from animals at indicated time points and stained for surface markers followed by fixation and permeabilization then intracellular staining using one of two panels: anti-CD3 (AF488), anti-FoxP3 (PE), anti-CD25 (PerCP-Cy5.5), anti-CD159a (PE-Cy7), anti-CD20 (APC) or anti-Granzyme B (Alexa Fluor 700), anti-CD4 (APC-H7), anti-Ki67 (BV421) or anti-CD45 (BV786), and anti-CD8 (BV510). Enumeration of the cell subsets depicted were determined using the phenotypic markers listed in Table S2. Samples were run on a BD FACSCanto II flow cytometer and analyzed using BD DIVA software at CRL.

Quantification and statistical analysis

Statistical significance was determined using GraphPad (Prism v.9.4.0), and significance was assigned at ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, and ∗∗∗∗p < 0.0001. Data are reported as mean ± SEM unless otherwise stated. For IL-2 and IL-10 stimulation of PBMC, in vivo IL-2 induced edema, tumor cell cytolysis, and non-human primate toxicology studies, significance was determined using ordinary one-way ANOVA followed by multiple comparison t-tests. A ratio pairwise t-test was used to determine significance in receptor blockade and tumor qPCR studies. One sample t-test was used to compare IRF induction. For evaluation of EBV muteins, TIL gene expression, rechallenge of cured mice, and human CD8⁺ T cell stimulation studies, Welch’s t-test was used. Mann-Whitney unpaired t-test was used in T cell cytokine assays with DK2¹⁰ (EGFR) treatment, FTY720 and rhIL-2 tumor efficacy studies, in vivo T cell phenotype and metabolism studies, and biodistribution studies. For ELISpot analysis, significance was determined with unpaired t-test. Statistical significance in in vitro metabolism studies was determined using one-way ANOVA followed by multiple comparison t-tests (OCR, ECAR) or Mann-Whitney test (spare respiratory capacity, ATP production rate). For T cell phenotype analysis with DK2¹⁰ (EGFR), IL-2, IL-10, and IL-2+IL-10 treatment, two-way ANOVA followed by multiple comparison t-tests was used. Statistical significance for tumor efficacy studies for scFv and targeted IL-2, IL-10, and DK2¹⁰ was determined by ordinary one-way ANOVA followed by multiple comparison t-tests.

Published: July 30, 2025