Potent antitumour activity of interleukin-2-Fc fusion proteins requires Fc-mediated depletion of regulatory T-cells

Interleukin-2 (IL-2) is an established therapeutic agent used for cancer immunotherapy. Since treatment efficacy is mediated by CD8+ and NK cell activity at the tumour site, considerable efforts have focused on generating variants that expand these subsets systemically, as exemplified by IL-2/antibody complexes and ‘superkines'. Here we describe a novel determinant of antitumour activity using fusion proteins consisting of IL-2 and the antibody fragment crystallizable (Fc) region. Generation of long-lived IL-2-Fc variants in which CD25 binding is abolished through mutation effectively prevents unwanted activation of CD25+ regulatory T-cells (Tregs) and results in strong expansion of CD25− cytotoxic subsets. Surprisingly, however, such variants are less effective than wild-type IL-2-Fc in mediating tumour rejection. Instead, we report that efficacy is crucially dependent on depletion of Tregs through Fc-mediated immune effector functions. Our results underpin an unexpected mechanism of action and provide important guidance for the development of next generation IL-2 therapeutics.


Generation of IL-2 mutants devoid of CD25 interactions
Single mutants were expressed as Fc-fusions in HEK293 cells, purified by affinity chromatography, and ranked based on their retained binding to bivalent human CD25 (hCD25) using biolayer interferometry ( Supplementary Figures 1B and   1C). This revealed that two of the substitutions, R38D and E61R, had the most pronounced effect on hCD25 binding. Surface plasmon resonance (SPR) measurements revealed that these mutations caused 20-70 fold reductions in affinity to monovalent hCD25 relative to the wild-type protein (Supplementary Figures 1D   and 1E). Moreover, their combination in a double mutant (IL-2 2X Fc) further reduced affinity to undetectable levels (K D >> 10 µM, Supplementary Fig. 1D). Consistent with these results, IL-2 R38D Fc, IL-2 E61R Fc and IL-2 2X Fc displayed nearly identical disruptions in binding to mouse CD25 ( Supplementary Fig. 1G). A similar profile was observed in in vitro assays using the CD25 high murine cell line CTLL-2, in which IL-2 2X Fc was the only variant to display complete abolition of binding and substantial reductions in proliferation and signalling (Supplementary Figures 1A and B). In contrast, introduced mutations only showed minimal reductions in activation of CD122 high cells (Supplementary Fig. 2C) and affinity to recombinant CD122 ( Supplementary Fig. 1F).
Assessment of in vivo activity revealed that IL-2 2X Fc mediated vigorous expansion of CD122 high memory-phenotype (MP) CD8 + T-cells and NK cells in the spleens of C57BL/6 mice, considerably higher than that observed after treatment with IL-2/mAb complexes, IL-2 WT Fc, IL-2 R38D Fc or IL-2 E61R Fc ( Supplementary Fig. 3A).
This activity was also superior to that of IL-2 F42A Fc ( Supplementary Fig. 3A), which incorporates a previously reported CD25-disrupting mutation 1 that we found was insufficient to abolish hCD25 binding ( Supplementary Figures 1E and 1F). The remarkable increase in the numbers of MP CD8 (30-fold) and NK cells (16-fold) after single dose IL-2 2X Fc administration resulted from a combination of increased subset frequencies ( Supplementary Fig. 3C), increased spleen lymphoid cellularity ( Supplementary Fig. 3B), induction of a major shift in the CD8 + to CD4 + T-cell ratio ( Supplementary Fig. 3D) and prolongation of serum half-life ( Supplementary Fig.   3E). Although a prolonged serum half-life was not anticipated, this is likely the result of abolished interaction with CD25, since this IL-2R subunit is known to participate in IL-2 homeostasis 8 . Half-life prolongation was further validated using an in vivo reporter assay, in which a single dose of IL-2 2X Fc given 48 h before adoptive transfer of MP CD8 cells was able to drive their proliferation ( Supplementary Fig. 3F).
Although the disruption of CD25 binding in the IL-2 2X Fc double mutant resulted in preferential expansion of MP CD8 and NK cells, this variant also induced a lower but considerable expansion of CD4 + CD25 + Tregs (5-fold, Supplementary   Fig. 3A). This might be explained by the ability of IL-2 2X Fc to induce residual levels of proliferation and signalling in CD25 high cells in vitro (Supplementary Fig. 2B).
Thus, to further improve selectivity, we generated triple mutants combining the identified R38D and E61R mutations with a range of additional substitutions.
Addition of the K43D mutation tested in our initial scan (Supplementary Figures 1B   and 1C) resulted in poor expression yields and low biological activity ( Supplementary   Fig. 4A), suggesting that a delicate structural balance needs to be maintained in order to prevent activity loss. Therefore, we generated a panel of triple mutants in order to 20 identify permissible substitutions at position 43 ( Supplementary Fig. 4B). In vivo screening in C57BL/6 mice led to the identification of a triple mutant with exquisite specificity towards CD122 high subsets, carrying the charge reversal substitution K43E ( Figure 1A and Supplementary Fig. 4C). This triple mutant (IL-2 3X Fc) displayed potent activity and an increased expansion ratio of MP CD8 and NK cells over Tregs, which remained at levels comparable to PBS control mice (Supplementary Figures 4D   and 4E). This remarkable selectivity resulted from further reductions in proliferation and activation of CD25 + cells, both in vitro ( Supplementary Figures 4H and 4I) and in vivo ( Supplementary Figures 4E-4G).
We next compared the cellular biodistribution profiles of fluorescently labelled IL-2 WT Fc, IL-2 2X Fc and the selected IL-2 3X Fc in the spleens of treated mice (Supplementary Figures 5A and 5B). Importantly, association with total CD4 + T-cells was reduced as the number of CD25-disrupting mutations increased (F Supplementary Fig. 5E). Since the majority of IL-2-Fc binding in the CD4 + compartment is mediated by Tregs ( Supplementary Fig. 5C), this suggests that IL-2 2X Fc and IL-2 3X Fc have a reduced ability to target this subset. Indeed, IL-2 WT Fc bound efficiently to Tregs (~ 82% IL-2-Fc + ), which was markedly reduced for IL-2 2X Fc (~ 8% IL-2-Fc + ) and abolished for IL-2 3X Fc (< 0.5% IL-2-Fc + ) ( Supplementary   Figures 5A, 5B and 5F). Although the IL-2 3X Fc triple mutant displayed reduced retention in the spleen (Supplementary Fig. 5C) it targeted more than 95% of NK cells in this organ, thus confirming its ability to interact with CD122 high CD25subsets (Supplementary Figures 5A and 5B).