The immune cytokine interleukin-2 (IL-2) is a potent anti-cancer molecule which stimulates the immune system to attack tumor cells, however, the toxic side effects limits its clinical usefulness. The mechanism of this toxicity is not well understood in humans. Interestingly, animals that lack the alpha subunit of the IL-2 receptor have reduced toxicity with IL-2 administration. Many efforts have been undertaken to modify IL-2’s interaction with this subunit in order to limit the toxicity in humans, however, this leads to reduced stability and potency and does not completely eliminate the interaction between the two. In addition, researchers have tried to alter IL-15 because the cytokine elicits similar signaling effects to IL-2 but has no affinity for the alpha subunit of the receptor. However, low stability of the native protein make this effort challenging.
In a recent study published in Nature, researchers at the University of Washington’s Institute for Protein Design (IPD) in Seattle used computational design to create a de novo protein mimetic of IL-2/IL-15 that has improved therapeutic properties without the toxic side effects. Using their computer program called Rosetta, the researchers designed from scratch a stable protein with binding sites mimicking those on IL-2 and IL-15 for all subunits of the IL-2 receptor except the alpha subunit. Compact proteins were designed to serve as scaffolds to hold the binding sites in the proper position and these designed scaffolds went through two rounds of screening to improve affinity and stability. Using yeast display, the top variants were selected for experimental characterization. The variants that bound the IL-2 receptor at low concentrations with the highest affinity were subjected to site-saturation mutagenesis and selected for increased affinity for the IL-2 receptor lacking the alpha subunit. A second round of selection on combinatorial libraries revealed the variants with the best combined individual affinity-enhancing substitutions for this receptor.
Interestingly, the designed protein, named Neo-2/15, had a higher stability and binding affinity for IL-2 receptors and greater stimulatory effect on natural killer cells and mouse primary T cells in vitro than naturally occurring IL-2. In mice, Neo-2/15 caused less expansion of the population of immunosuppressive T regulatory (Tregs) cells than native IL-2, leading to a higher ratio of killer T cells to regulatory T cells. Treatment with Neo-2/15 caused a dose-dependent delay in tumor growth in both a colon cancer and a melanoma mouse model. In addition, mice treated with a combination of Neo-2/15 and the anti-TRP1 monoclonal antibody, TA99, showed significantly reduced toxicity and a superior therapeutic effect compared to mouse IL-2.
This study provides strong evidence that computational design of proteins from scratch can lead to bio-superior molecules with enhanced therapeutic properties and lesser side effects thus opening the doors for new approaches in the discovery of protein-based therapeutics.