Investigating plasticity within the interleukin-6 family with AlphaFold-Multimer

Abstract

Cytokines are important soluble mediators that are involved in physiological and pathophysiological processes. Among them, members of the interleukin-6 (IL-6) family of cytokines have gained remarkable attention, because especially the name-giving cytokine IL-6 has been shown to be an excellent target to treat inflammatory and autoimmune diseases. The IL-6 family consists of nine members, which activate their target cells via combinations of non-signaling α- and/or signal-transducing β-receptors. While some receptor combinations are exclusively used by a single cytokine, other cytokine receptor combinations are used by multiple cytokines. Research in recent years unraveled another level of complexity: several cytokine cannot only signal via their canonical receptors, but can bind to and signal via additional α- and/or β-receptors, albeit with less affinity. While several examples of such cytokine plasticity have been reported, a systematic analysis of this phenomenon is lacking. The development of artificial intelligence programs like AlphaFold allows the computational analysis of protein complexes in a systematic manner. Here, we develop a analysis pipeline for cytokine:cytokine receptor interaction and show that AlphaFold-Multimer correctly predicts the canonical ligands of the IL-6 family. However, AlphaFold-Multimer does not provide sufficient insight to conclusively predict alternative, low-affinity ligands for receptors within the IL-6 family.