WRAP, noun: a custom-built protein jacket that covers a membrane protein’s greasy outside so it can survive in water; except Mihaljević and teammates found the jacket can also preserve the protein’s real shape, binding sites, active machinery, and vaccine-relevant surfaces while detergents watch from the bench.

First Quarter: The Slippery Opponent

Membrane proteins are the all-star gatekeepers of biology. They sit in cell membranes, move nutrients, sense signals, and give drug developers that “we might have something here” look. They also make up roughly a third of the proteome and more than half of approved drug targets, which is a scouting report you cannot ignore [1].

The problem? They hate water. Their outside surfaces are hydrophobic because, in their natural habitat, they snuggle up against fatty membranes. Pull them out, and they clump, collapse, or demand detergents like a superstar demanding a private locker room. Detergents can help, but they also complicate structural biology, drug screening, and vaccine work. Sometimes they keep the player upright. Sometimes they change the play.

Second Quarter: AI Draws Up a Trick Play

The WRAP idea is wonderfully direct: do not mutate the membrane protein into a weird water-friendly cousin. Instead, design a new protein that wraps around the oily parts, with a nonpolar inner face gripping the target and a water-friendly outer face facing the crowd.

That design pipeline uses the modern protein-AI bench. RFdiffusion helps generate protein backbones, ProteinMPNN helps choose amino acid sequences, and AlphaFold-style structure prediction helps screen whether the design looks likely to fold as intended [2-5]. If that sounds like a lot of computational choreography, it is. The GPUs are doing wind sprints.

The key move is that WRAPs are genetically encoded. Researchers can express the WRAP and target together in E. coli, then purify the complex from the soluble fraction. Translation: the membrane protein may never need to go into the membrane at all. That is not just a clever play. That is stealing second while the pitcher is checking the scoreboard.

Third Quarter: The Scoreboard Lights Up

The team tested WRAPs across several classes: beta-barrel outer membrane proteins, helical multipass proteins, monomers, and oligomers. In the preprint version, WRAPed E. coli OmpA stayed soluble and antibody-recognizable, while WRAPed GlpG, a membrane protease, retained activity in an enzyme-labeling assay [2]. That matters because a pretty structure without function is just molecular cosplay.

Then came the structural buzzer-beater: the published Science paper reports a 2.95 angstrom cryo-EM structure of a WRAPed mycobacterial porin [1]. Porins are channel proteins, basically cellular doorways with security staff. Getting a high-resolution structure in solution suggests the WRAP did not just duct-tape the protein into some awkward pose. It preserved enough of the native architecture to make structural biology useful.

The team also generated soluble versions of Treponema pallidum antigens, aiming straight at a long-running obstacle in syphilis vaccine research [1,2]. Those outer membrane proteins have been hard to produce and characterize. WRAPs give researchers a cleaner shot at studying them, generating antibodies, and asking whether they can work as vaccine components. The trophy is not being handed out yet, but the underdog just hit a three.

Fourth Quarter: Why This Could Matter

If WRAPs generalize, they could make membrane proteins less miserable to study. That would help drug discovery teams screen ligands against cleaner soluble targets. It could help structural biologists solve proteins that normally require fussy detergent conditions. It could help vaccine researchers present antigens closer to their real shape without rewriting the native sequence.

The field around this is moving fast. RFdiffusion showed that diffusion models can design new protein structures and functions [3]. AlphaFold 3 expanded structure prediction toward biomolecular interactions [5]. Recent work has also pushed de novo membrane channel and beta-barrel design forward [6,7]. WRAPs sit right in that championship bracket: not just predicting biology, but building useful molecular hardware.

Still, keep the champagne corked. WRAPs need target structures or strong predictions. They still need experimental validation. A WRAP might hide or alter some surfaces, depending on the antigen or assay. And vaccine relevance is a long season, not a highlight reel. Soluble antigens are a start, not a licensed shot in the arm.

But as a play design, this one has style. Instead of forcing membrane proteins to stop being themselves, WRAPs give them a custom arena where they can behave. Biology’s greasiest players finally get traction, and the crowd, by which I mean structural biologists with too much coffee, has every reason to lean forward.

References

1. Mihaljević L. et al. “Membrane protein solubilization and structure determination using de novo-designed proteins.” Science 393(6806):eadr3817, 2026. DOI: 10.1126/science.adr3817. PMID: 42391386.

2. Mihaljević L. et al. “Solubilization of Membrane Proteins using designed protein WRAPS.” bioRxiv, 2025. DOI: 10.1101/2025.02.04.636539. PMCID: PMC11838538.

3. Watson J.L. et al. “De novo design of protein structure and function with RFdiffusion.” Nature 620, 1089-1100, 2023. DOI: 10.1038/s41586-023-06415-8.

4. Dauparas J. et al. “Robust deep learning-based protein sequence design using ProteinMPNN.” Science 378, 49-56, 2022. DOI: 10.1126/science.add2187.

5. Abramson J. et al. “Accurate structure prediction of biomolecular interactions with AlphaFold 3.” Nature 630, 493-500, 2024. DOI: 10.1038/s41586-024-07487-w.

6. Kim D.E. et al. “Parametrically guided design of beta barrels and transmembrane nanopores using deep learning.” PNAS 122(38):e2425459122, 2025. DOI: 10.1073/pnas.2425459122.

7. Li Y. et al. “Water, Solute, and Ion Transport in De Novo-Designed Membrane Protein Channels.” ACS Nano 19(2), 2185-2195, 2025. DOI: 10.1021/acsnano.4c11317.

Disclaimer: This blog post is a simplified summary of published research for educational purposes. The accompanying illustration is artistic and does not depict actual model architectures, data, or experimental results. Always refer to the original paper for technical details.