De novo-designed transmembrane proteins bind and regulate a cytokine receptor

Marco Mravic, Li He, Huong T. Kratochvil, Hailin Hu, Sarah E. Nick, Weiya Bai, Anne P. B. Edwards, Hyunil Jo, Yibing Wu, Daniel DiMaio, William F. DeGrado

TL;DR

Researchers have successfully designed de novo transmembrane proteins that target and inhibit the erythropoietin receptor (EpoR) by binding its membrane-spanning region in a custom, non-natural orientation. This breakthrough demonstrates that we can now computationally engineer precise "molecular wrenches" to regulate cell signaling from within the lipid bilayer, moving beyond simply mimicking nature.

What Problem Does This Paper Address?

Membrane proteins are the gatekeepers of the cell, serving as the targets for over 50% of modern drugs. However, most drugs target the "water-exposed" parts of these proteins—the bits sticking out into the extracellular or intracellular space. The transmembrane (TM) domains, buried deep within the oily lipid bilayer, remain a "dark matter" for drug design.

While the TM regions are often critical for a receptor’s function (like the dimerization required for signaling), they are notoriously difficult to target. Previous efforts have mostly relied on "mimicry"—creating peptides that copy a receptor's own natural binding motifs to gum up the works. The challenge addressed here is whether we can use computational design to create a protein from scratch (de novo) that binds a TM domain in a completely custom geometry to precisely regulate biological function.

Methods

The authors employed an upgraded version of the CHAMP (Computed Helical Anti-Membrane Protein) algorithm. Their workflow involved several sophisticated steps:

  • Structural Informatics: Instead of copying the EpoR's natural parallel binding mode, the team mined a database of natural TM interactions to find a stable "antiparallel" geometry. They identified a "small-X6-small" motif (where small amino acids like Serine or Glycine allow helices to pack tightly) as a promising target site.
  • Computational Design (RosettaMP): Using the Rosetta software suite adapted for membrane environments, they modeled an idealized antiparallel backbone and optimized the side-chain sequences to maximize binding affinity and packing against the EpoR TM domain.
  • Automated Selection: They moved away from subjective human visual inspection, instead using automated "PackStat" scores to rank designs based on the absence of packing voids at the interface.
  • Multimodal Validation: The designs were validated using:
  • Thiol-disulfide exchange: To measure competitive binding stability in model membranes.
  • FRET & Split-GFP: To confirm the 1:1 stoichiometry and the intended antiparallel orientation in live cells.
  • Solution NMR: To map the exact atoms involved in the binding interface.

Datasets

The study relied on:

  • Structural Databases: A previously published database of globally clustered TM helix-helix interaction geometries (specifically "Cluster 6" of antiparallel helices) was used to generate the design templates.
  • PDB Structures: The "no-design" control was derived from the Photosystem II light-harvesting complex (PDB: 3BZ1).
  • Experimental Benchmarks: The designs were tested against wild-type mouse EpoR, human EpoR, and various chimeric and mutant versions to prove specificity.

Key Insights and Ideas

  • Topology is a Control Knob: The EpoR receptor normally signals by forming a parallel homodimer. By designing a protein that binds in an antiparallel orientation, the researchers created a "dominant-negative" inhibitor that physically prevents the receptor from adopting its active state.
  • Exquisite Specificity: The designed proteins (CHAMP-1 and CHAMP-2) were remarkably specific. They could distinguish between mouse and human EpoR, which differ by only three amino acids in their TM regions.
  • The "Malleable" Surface: The study proves that TM domains aren't locked into one "correct" binding mode. They present a malleable molecular surface that can be targeted via different, non-natural geometries if the binder is designed with high enough complementarity.
  • Automation of Design: By formalizing the rules for backbone selection and sequence ranking, the authors have made de novo TM design more reproducible and accessible to the broader scientific community.

Key Results

  • Functional Inhibition: In mammalian BaF3 cells, the expression of CHAMP proteins reduced erythropoietin-induced cell proliferation by approximately 60%.
  • Signaling Blockade: Immunoblots confirmed that CHAMP-1 reduced the phosphorylation of JAK2 and STAT5, the primary downstream effectors of EpoR signaling.
  • Structural Accuracy: Solution NMR and FRET data confirmed that the complex formed a stable 1:1 heterodimer. The NMR chemical shift perturbations aligned closely with the predicted interface in the computational model, particularly around the targeted "small-X6-small" motif.
  • Tunability: Using a titratable promoter, the researchers showed that the level of inhibition could be precisely tuned by controlling the concentration of the CHAMP protein.

Limitations and Future Work

While a major step forward, the authors note some limitations:

  • Environmental Sensitivity: There were slight discrepancies between the binding interfaces observed in detergent micelles (NMR) versus those inferred from activity in live cell membranes. This suggests that the local lipid environment still influences the fine details of TM packing.
  • Scope: This study focused on a single-pass TM receptor. Designing binders for multi-spanning membrane proteins (like GPCRs or ion channels) will be significantly more complex.
  • Pharmacology: As small, hydrophobic proteins, delivering these as "drugs" remains a challenge. Future work may look at expressing them via gene therapy or developing more drug-like small molecule mimetics based on these design principles.

Reference: Mravic, M., He, L., et al. (2024). De novo-designed transmembrane proteins bind and regulate a cytokine receptor. Nature Chemical Biology, 20, 751–760. https://doi.org/10.1038/s41589-024-01562-z

Summary updated

Abstract

Transmembrane (TM) domains as simple as a single span can perform complex biological functions using entirely lipid-embedded chemical features. Computational design has the potential to generate custom tool molecules directly targeting membrane proteins at their functional TM regions. Thus far, designed TM domain-targeting agents have been limited to mimicking the binding modes and motifs of natural TM interaction partners. Here, we demonstrate the design of de novo TM proteins targeting the erythropoietin receptor (EpoR) TM domain in a custom binding topology competitive with receptor homodimerization. The TM proteins expressed in mammalian cells complex with EpoR and inhibit erythropoietin-induced cell proliferation. In vitro, the synthetic TM domain complex outcompetes EpoR homodimerization. Structural characterization reveals that the complex involves the intended amino acids and agrees with our designed molecular model of antiparallel TM helices at 1:1 stoichiometry. Thus, membrane protein TM regions can now be targeted in custom-designed topologies.
BibTeX
@inproceedings{mravic2024de,
  author = {Marco Mravic and Li He and Huong T. Kratochvil and Hailin Hu and Sarah E. Nick and Weiya Bai and Anne P. B. Edwards and Hyunil Jo and Yibing Wu and Daniel DiMaio and William F. DeGrado},
  title = {{De novo-designed transmembrane proteins bind and regulate a cytokine receptor}},
  booktitle = {Nature Chemical Biology},
  year = {2024},
  doi = {10.1038/s41589-024-01562-z},
}
Figure