Membrane proteins form dynamic, often transient contacts with surrounding lipids and with small-molecule or carbohydrate ligands. Capturing these concomitant interactions is technically challenging. The authors report that generative deep learning–designed WRAP domains can be fused to membrane proteins to preserve weak lipid and ligand interactions during native mass spectrometry analysis. Using WRAP-fused GlpG, AqpM, and OmpA as model systems, the approach retains and enables characterization of lipid interactions without detergent micelles. In the WRAP-OmpA construct, selective binding of phosphatidylethanolamine lipids was observed in cavities at the protein–WRAP interface, alongside weak chitobiose ligand binding. The study positions WRAPs as versatile tools to probe ligand and lipid interactions of membrane proteins.
The authors applied generative deep learning to design WRAP domains intended to stabilize membrane proteins and maintain their native, weak interactions during gas-phase analysis by mass spectrometry. The key concept is that WRAP fusions can shield or scaffold membrane proteins so that associated lipids and small ligands remain bound in the absence of detergent micelles, allowing these interactions to be detected and characterized by native MS.
The source reports that WRAPs are effective at retaining lipid interactions that are typically lost when membrane proteins are extracted from lipid environments or when detergents are removed. This preservation enables the simultaneous observation of both lipid and ligand binding events in a single experimental workflow.
Three membrane proteins were used as model systems to evaluate the WRAP-fusion strategy: GlpG, AqpM, and OmpA. Each protein was generated as a WRAP-fused construct and analyzed by native mass spectrometry to assess whether lipid and ligand interactions were retained in the absence of detergent micelles.
Although the source does not provide detailed experimental parameters in the summary, it emphasizes that the WRAP fusions enabled retention and detection of weak, transient contacts that are otherwise difficult to capture during native MS. The approach leverages the structural and stabilizing properties of the designed WRAP domains to present membrane proteins in a form amenable to gas-phase analysis while maintaining associated molecules.
A principal result highlighted in the source concerns the WRAP-OmpA construct. The authors show that WRAP-OmpA selectively binds phosphatidylethanolamine lipids within cavities formed at the interface between the membrane protein and the WRAP domain. This selective lipid binding was observed without detergent micelles present during mass spectrometry.
Concurrently, the WRAP-OmpA construct was able to accommodate weak binding of a chitobiose ligand, demonstrating that the WRAP approach can preserve both lipid and small-ligand interactions simultaneously. The source frames this as evidence that WRAPs can scaffold membrane proteins in a conformation that retains biologically relevant, but weak, interactions for native MS characterization.
The findings establish designed WRAP domains as adaptable vehicles for studying membrane protein interactions by native mass spectrometry. Potential implications reported in the source include:
Improved ability to detect and characterize transient or weak protein–lipid interactions that are important for membrane protein structure and function.
Simultaneous observation of lipid binding and small-molecule or carbohydrate ligand interactions on the same membrane protein construct, providing a more complete view of the protein’s microenvironment.
Use of WRAP fusions to avoid detergent micelles during analysis, which can simplify interpretation of native MS data and reduce artefacts introduced by detergents.
The source presents these outcomes as proof-of-concept evidence that de novo designed WRAP domains can extend the capabilities of native mass spectrometry for membrane protein research. Details on broader validation, limits of the approach, or comparative performance metrics relative to alternative strategies were not reported in the source summary.
The authors declared no competing interests in the source. Funding sources listed include the Swedish Research Council, Swedish Cancer Society, Swedish Society for Medical Research, Knut and Alice Wallenberg Foundation, and Karolinska Institutet. The article is a preprint posted on bioRxiv and has not been peer reviewed.