Membrane proteins are fundamental to a myriad of biological processes, serving as vital channels, molecular sensors, and prominent drug targets. Their structural complexity is compounded by the requirement to navigate the hydrophobic landscape of cellular lipid bilayers, a process that is often fraught with misfolding and aggregation. Designing these proteins from first principles has historically been a formidable challenge. Unlike soluble proteins whose folding pathways are relatively well understood, membrane proteins undergo a more nuanced folding journey influenced by the lipid environment, necessitating novel conceptual frameworks beyond the traditional pursuit of maximal thermodynamic stability.

Historically, the prevailing dogma in protein engineering has been to craft proteins with the highest possible stability in their final conformational state, assuming that this would naturally encourage correct folding and functional assembly. However, the recent study overturns this assumption specifically for transmembrane β-barrel proteins. Using an innovative cell-free expression system augmented with synthetic lipid vesicles, the research team demonstrated that designs optimized solely for maximal stability paradoxically exhibited a propensity for premature folding in aqueous pre-membrane environments. This premature folding predisposed the proteins to self-associate into aggregates rather than successfully inserting into membranes.

Giacomo Pedrelli, the study’s lead author and a PhD candidate at VIB-VUB, explains, “Focusing exclusively on final-state stability can inadvertently trap proteins in off-pathway intermediates. When these proteins fold prematurely in the aqueous phase, they tend to aggregate, thwarting their journey to proper membrane insertion.” This insight underscores a vital caveat in membrane protein design: the folding pathway and its kinetics are as crucial as the stability of the endpoint.

To circumvent this obstacle, the researchers adopted a strategy termed negative design, which involves the deliberate introduction of destabilizing features or ‘imperfections’ in the protein sequence that hinder premature folding. By subtly reducing stability at strategic positions, negative design ensures that the protein remains sufficiently dynamic during its nascent stage to avoid aggregation and retain the correct folding trajectory. Remarkably, these engineered destabilizations did not compromise the ultimate stability of the protein once properly inserted into the lipid bilayer; rather, they facilitated an efficient folding pathway culminating in correctly assembled functional proteins.

Beyond the experimental manipulations, the study leveraged cutting-edge computational tools to identify optimal sites for negative design mutations. Notably, a sophisticated protein language model, ESM3, trained on extensive evolutionary data, outperformed traditional physics-based algorithms in predicting beneficial destabilizing mutations. While conventional methods flagged these mutations as largely deleterious, the AI-driven model accurately discerned changes that enhanced membrane assembly, highlighting the complementary power of evolutionary-informed machine learning approaches in guiding synthetic biology efforts.

This advancement marks a critical step forward in the rational design of transmembrane β-barrel proteins, structures that form barrel-shaped pores traversing membranes. These nanopores have vast potential applications, including ultrasensitive biosensing, molecular filtration, and as pivotal components in next-generation DNA and RNA sequencing technologies. Engineering such pores with high precision and reliability has, until now, been hindered by difficulties in controlling folding and membrane integration.

Professor Anastassia Vorobieva, senior researcher and co-author, emphasizes the paradigm shift ushered in by these findings: “Our results highlight that static considerations of a protein’s folded structure are insufficient. We must holistically consider the thermodynamic and kinetic landscape of the entire folding journey. Incorporating negative design principles enables us to steer proteins along productive folding routes, unlocking novel functional possibilities.”

The implications of this research extend well beyond academic curiosity. By enabling stable yet foldable transmembrane proteins in artificial membranes, this approach could accelerate the development of tailored synthetic proteins for use in biotechnology, nanotechnology, and medicine. For example, custom-designed nanopores could revolutionize molecular diagnostics, environmental sensing, and targeted drug delivery, while engineered membrane proteins may serve as novel therapeutic targets or biocompatible devices.

This study showcases how a nuanced balance between stability and controlled instability—guided by advanced computational models and precise experimental methods—can overcome fundamental bottlenecks in membrane protein design. This novel negative design framework challenges dogmatic strategies and invites a rethinking of protein engineering principles to embrace dynamic folding pathways as critical determinants of functional success.

As membrane protein design continues to evolve, integrating machine learning with experimental innovation promises to unlock unprecedented possibilities in synthetic biology. The work of the VIB–VUB team stands as a clarion call for interdisciplinary approaches that marry theoretical insight, computational power, and biochemical ingenuity to solve enduring challenges in biomolecular engineering.

With this holistic strategy, the once elusive goal of reliable, efficient, and scalable design of functional transmembrane β-barrels appears increasingly within reach, catalyzing new horizons in the manufacture and application of synthetic membrane proteins.

Subject of Research: Not applicable

Article Title: Negative design enables cell-free expression and folding of designed transmembrane β-barrels

News Publication Date: 14 April 2026

Web References: http://dx.doi.org/10.1073/pnas.2528772123

References: Proceedings of the National Academy of Sciences, 8-Apr-2026, DOI: 10.1073/pnas.2528772123

Keywords: Molecular biology, Biochemistry, Biophysics, Computational biology, Signal transduction