Modulating G-CSFR Signaling Using De Novo Designed Cytokines

Tightly modulating cytokine receptors, which naturally decode complex patterns of downstream signaling, is key to treating a wide range of diseases. De novo protein design is a powerful approach for creating ligands that can modulate the receptor into a range of distinct states, steering diverse functional outcomes. Here, we first aimed to tackle the granulocyte colony-stimulating factor receptor (G-CSFR) by designing binders with idealized structural properties using our in-house protein design software. We then tailored these binders to associate with G-CSFR in different configurations, with the three-pronged aim of developing full agonists, tuning agonists, or potent inhibitors. Unlike the G-CSF, our designed binders were easy and cheap to make, small in size, hyper thermostable, and highly resistant to proteolysis. In addition, NMR and X-ray crystallography showed that the structures of the binders match the designed models to atomic accuracy. Through tandem fusion and affinity maturation of our designed binders, we obtained proteins that potently bind and activate G-CSFR at picomolar levels. These agonists possessed strong granulopoietic activity in vitro and in vivo. Such molecules could be clinically useful as granulopoietic agents, especially given their superior production and stability properties compared to recombinant human G-CSF (rhG-CSF).

Furthermore, by deriving rigidly fused binders, we sought to create tuning agonists that dimerize G-CSFR in non-natural orientations. This was necessary in order to reduce the pleiotropic activities of native G-CSF, which is known to modulate several cellular functions, including proliferation, myeloid differentiation of hematopoietic stem cells, mobilization of these cells and neutrophils from the bone marrow, protection from apoptosis and activation of various immune cell subsets. Single-molecule imaging and cellular assays indicated that designed orientation-rigging ligands differentially associate with and activate G-CSFR, which was further confirmed by assessment of G-CSFR downstream signaling profiles. Design molecules modulated a subset of G-CSF-controlled genes, showing a clear bias towards genetic programs that activate granulocytic differentiation over those that regulate HSPC proliferation or immune cell activation, and thus had reduced functional pleiotropy, highlighting their therapeutic potential for induction of granulopoiesis.

Finally, we also sought to engineer binders that block G-CSFR, which we achieved by designing monomeric variants that disfavor receptor dimerization in the plasma membrane environment. These inhibitors bound the receptor more tightly than G-CSF, and outcompeted it in competitive binding assays at low nanomolar levels. These inhibitors also offer promising therapies for leukemic and autoinflammatory disorders in which G-CSFR signaling plays an importnat role. These non-activating ligands could also be used as homing fusions to guide payloads to G-CSFR-expressing cells.

In summary, de novo protein design has successfully generated therapeutic proteins with unique G-CSFR binding activities that can be further explored for clinical use. A similar approach can be used to modulate the activities of several other cytokines.