Bottom-up protein assemblies form efficient RNA transport systems

Researchers have built functional RNA transfer systems from scratch using synthetic protein assemblies, mimicking the efficiency viruses developed over billions of years. The team created more than 100 bottom-up RNA vehicles with unique shapes and sizes, surpassing the RNA transfer efficiency of widely used delivery vehicles by several orders of magnitude.

This approach combines natural protein domains with artificial intelligence-designed protein architectures, offering a new way to construct nucleic acid transport independent of evolutionary constraints. The work demonstrates the in vivo biodistribution of one of these vehicles in cleared mice and their potential for gene-editing treatment strategies in patient-derived cells and a pig.

AI-Designed Protein Assemblies Create Novel RNA Vehicles

Artificial intelligence designed protein structures that bypass natural limitations, enabling the creation of RNA delivery vehicles. The team discovered that STVs built on cyclic symmetries transferred more RNA per packaging protein than those with icosahedral symmetry. High delivery efficiency was confirmed with EGFP mRNA into unmodified cells using four designs, HE0490, HE0499, HE0505 (all D3 symmetry) and HE0690 (C8 symmetry).

This suggests that virtually limitless numbers of synthetic RNA delivery vehicles can be created by harnessing artificial intelligence and non-natural symmetries. “STVs are distinct from known natural RNA transfer vehicles, exhibiting unique characteristics that include cyclic and dihedral symmetries, open structures and low complexity of the assembled protein,” the paper states.

Beyond efficiency, the research demonstrates programmable tropism; the vehicles’ targeting ability can be controlled by incorporating computationally designed peptide binders. This allows for delivery of therapeutically relevant RNA cargo into a diverse range of cellular models. In vivo biodistribution of one STV was observed in a mouse at near-single-cell resolution, confirming its safety and enabling a gene-editing treatment strategy for Duchenne muscular dystrophy in both patient-derived cells and a pig.

Natural Virus Efficiency Compared to Synthetic Transfer Vehicles

Protein assemblies created using artificial intelligence now demonstrate RNA transfer efficiency exceeding that of natural systems by several orders of magnitude, a finding that re-evaluates the long-held assumption that billions of years of evolution are required for optimal nucleic acid delivery. The research reveals that specifically designed, bottom-up RNA transfer vehicles, dubbed synthetic transfer vehicles or STVs, can surpass the performance of established biological and chemical delivery methods.

This leap in efficiency stems from the ability to circumvent evolutionary constraints and explore protein architectures absent in nature, allowing for optimization beyond what is typically observed in naturally occurring systems. The team’s approach focused on constructing STVs from natural protein domains combined with AI-designed synthetic protein assemblies, resulting in more than 100 unique designs with varying shapes and sizes.

These designs were then screened for RNA transfer capability, revealing that certain symmetries, cyclic and dihedral, proved particularly effective at maximizing RNA packaging per protein unit. The resulting STV-C8 architecture not only outperformed existing methods but also proved versatile enough to deliver diverse RNA cargoes, including transcription factors and gene editors, into human, mouse, and pig cells. Beyond sheer efficiency, the research demonstrates a level of programmability previously unseen in RNA delivery systems.

Incorporation of computationally designed peptide binders allows for targeted delivery of therapeutic RNAs into specific cellular models, expanding the potential applications of these STVs. This programmable tropism, coupled with the ability to create a large number of STV designs, suggests a future where RNA-based therapies can be tailored to individual patients and diseases with increased precision. The work establishes a multidimensional screening platform capable of testing hundreds of STV constructs.

STV-C8 Structure: Planar Symmetry & RNA Packaging

STV-C8 distinguishes itself through an unusual planar symmetry, identified as the most effective structure for delivering RNA via a multidimensional screening system testing more than 100 designs. Characterization of STV-C8 revealed details about its shape, RNA content, and packaging capacity, with researchers programming its targeting ability by integrating computationally designed peptide binders. Analyses demonstrated STV-C8 achieves RNA transfer efficiency several orders of magnitude greater than natural systems and lipid nanoparticles (LNPs).

To explore the potential for scaling production, the team designed 30 additional assemblies sharing the C8 symmetry. Despite utilizing the same symmetry, these new structures exhibited considerable diversity in both sequence and structural composition, allowing for screening based on specific characteristics like altered packaging density or reduced immunogenicity. Further expanding the design possibilities, the researchers combined the assembly domain with a diverse panel of membrane-binding domains identified through structure-based searches using FoldSeek.

The team selected 29 membrane-binding domains from unrestricted searches across all species, restricted searches for human proteins, and metagenomic proteins within the ESMAtlas. Testing a library of these domains fused to SynL-tdPCP-HE0690 revealed that the membrane-binding domain from Ursus americanus (Ua PHPLC) was particularly efficient for RNA packaging and transfer. Validation confirmed robust membrane localization of Ua PHPLC-STVs in producer cells and their ability to efficiently transfer EGFP mRNA into target cells.

The resulting construct, combining Ua PHPLC and SynL-tdPCP-HE0690, was named STV-C8: Characterization and programming of the STV-C8 RNA carrier. “Viruses are highly diverse, yet most have converged towards packaging their genomes in large, multimeric protein shells with icosahedral and helical symmetry,” the researchers write, adding that their approach “created atypical but functional RNA transport vehicles that outperformed their natural counterparts.” The high transport efficiency of STV-C8, coupled with its unique size and shape, suggests that circumventing evolutionary constraints can be advantageous when constructing RNA transport vehicles de novo.

Multidimensional Screening Identifies Efficient STV-C8 Design

A multidimensional screening system enabled the identification of STV-C8, a uniquely efficient structure for RNA delivery built upon an unusual planar symmetry. This performance was achieved through a systematic approach, evaluating designs across multiple parameters to optimize function. Further characterization revealed STV-C8’s capacity to package and deliver a variety of RNA cargo, including transcription factors, gene editors, and programmable antivirals, into human, mouse, and pig-derived cells and animal models.

Analysis of producer cells and STV-C8 using RNA sequencing showed strong correlations, indicating effective RNA encapsulation and preservation during transfer. Flow cytometry confirmed efficient delivery of EGFP mRNA into HEK293T cells, validating the system’s functionality in cellular models.

These variations were generated using RFdiffusion, a protein design tool, demonstrating the potential for creating a library of customized RNA delivery vehicles. “We demonstrated that protein assemblies designed by RFdiffusion can be harnessed to build synthetic RNA transfer vehicles from scratch,” the study reports, highlighting the potential for entirely new approaches to gene therapy and beyond. Both dimensions were measured as total protein content, with each data point representing the mean from six biological replicates.

Tropism Programming via Computationally Designed Peptide Binders

Tropism programming relies on computationally designed peptide binders fused to the synthetic RNA transfer vehicles, allowing for targeted delivery beyond natural limitations. Researchers demonstrated cell-type specificity by engineering STV-C8 with binders against EGFR or IL-7Rα, effectively repurposing the vehicle for distinct cellular targets without relying on antibodies. These binders, combined with a modified VSV-G protein lacking LDLR binding, guide the STV-C8 particles to specific receptors on cell surfaces, opening avenues for precision RNA delivery.

Flow cytometry analysis quantified Annexin V staining intensity, revealing that engineered STV-C8 particles successfully localized to cells expressing the target receptors. The process involved expressing minibinders, EGFRn, EGFRc, and IL-7Rα, as fusion constructs alongside STV-C8 components and the VSV-G mutant in producer cells. This approach bypasses the need for antibodies, addressing a limitation in targeting options and potentially reducing immunogenicity.

The team also tested a CD19 scFv binder, further demonstrating the versatility of this tropism programming method. The design of these peptide binders utilizes artificial intelligence, enabling the creation of proteins with tailored binding affinities and specificities. By fusing the binders to a signal peptide and transmembrane domain, researchers ensured their proper display on the STV-C8 surface. This modular design allows for rapid adaptation of the vehicle’s targeting profile, potentially facilitating the delivery of RNA-based therapeutics to a wide range of cell types.

“We considered whether the concept of using artificially designed proteins could be extended to the programming of cell-type specificity,” reflecting the team’s focus on overcoming limitations of existing delivery systems. The ability to engineer tropism through computationally designed binders represents a step towards realizing the full potential of synthetic RNA transfer vehicles.

STV-C8 Demonstrates Superior RNA Transfer Efficiency

STV-C8 consistently achieved higher expression of delivered EGFP mRNA in target cells when compared to established delivery systems like virus-like particles, enveloped protein nanocages, and selective endogenous encapsidation, according to recent findings. Researchers benchmarked STV-C8’s performance against these genetically encoded systems, alongside clinically utilized lipid nanoparticles, to quantify its RNA transfer capabilities. Delivering mRNA encoding Cre recombinase also showed superior efficiency with STV-C8, as demonstrated through Addgene-sourced plasmids and subsequent analysis.

The team established a multidimensional screening platform, enabling the testing of more than 100 STV constructs with diverse structures to optimize RNA delivery. This rigorous process led to the development of the STV-C8 architecture, which outperformed both biological and chemical vehicles in transferring a variety of cargo RNAs. Observations suggest STV-C8 may package cargo RNA on the surface of its oligomers, with the tdPCP RNA-binding protein positioned outside the HE0690 structure and the surrounding membrane providing protection.

Efficiency comparisons between STV-C8 and lipid nanoparticles were conducted on two levels; first, researchers measured the amount of mRNA needed to achieve the same percentage of transfected cells, titrating both STV and LNPs on target cells and analyzing EGFP expression via flow cytometry. Second, the expression levels induced by each vehicle type were directly compared, with STV-C8 surpassing lipid nanoparticles by several orders of magnitude. This work establishes a foundation for programmable delivery systems with potential applications extending beyond current biological and chemical limitations.

In Vivo Biodistribution and Safety Profile of STV-C8

Detailed analysis in a mouse model revealed near-single-cell resolution of STV-C8 biodistribution, demonstrating the vehicle’s capacity to distribute throughout the body following intravenous injection. Light-sheet microscopy of cleared mice showed amplified expression of EGFP delivered via STV-C8, contrasting with images from mice treated with empty vesicles, which exhibited minimal signal. These observations, captured 72 hours post-injection, provide visual confirmation of successful in vivo delivery and expression of the reporter RNA.

The scale bars in the images indicate a 5 millimeter measurement, contextualizing the extent of distribution observed. The safety profile of STV-C8 was further evaluated across two animal models, confirming a lack of adverse effects during the observation period. Beyond safety, the research team assessed STV-C8’s translational potential by employing CRISPR-Cas9 gene editing to address Duchenne muscular dystrophy (DMD) in both patient-derived and pig skeletal muscle cells.

Successful deletion of exon 51, a common target in DMD therapy, was confirmed, restoring the DMD open-reading frame in treated cells. This targeted exon deletion represents a functional correction of the genetic defect underlying the disease. While acknowledging that direct intramuscular delivery may not fully translate to DMD treatment strategies, the researchers noted, outlining next steps for refining the delivery system. However, the combination of a beneficial safety profile and demonstrated gene editing efficacy in relevant cellular and animal models positions STV-C8 as a strong candidate for further development as a therapeutic delivery vehicle.

STV-C8 Delivers CRISPR-Cas9 for Duchenne Muscular Dystrophy

This high-resolution imaging, achieved through whole-mouse clearing techniques combined with deep learning analysis, allowed researchers to trace the vehicle’s path and confirm its distribution throughout the animal’s body. The team extended this analysis to assess safety, finding no immunological or toxicological side effects following systemic injection of STV-C8 in two animal models, a critical step toward potential clinical translation. This approach builds on prior work demonstrating phenotypic rescue in a DMD pig model using CRISPR-Cas9, but seeks to circumvent potential severe side effects associated with adeno-associated viruses (AAVs) previously employed for delivery.

The versatility of STV-C8 extends beyond DMD, as the vehicle successfully transported diverse RNA cargo including reporter RNAs, gene editors, programmable antivirals, and transcription factors into cells from multiple species. “We demonstrate the versatility of STV-C8 by delivering various cargo RNAs,” the researchers report, highlighting the potential for broad application of this bottom-up RNA transfer system. Further studies detailed in Nature demonstrate the potential of this approach.

Membrane Bending Mechanism for Bottom-Up Vehicle Assembly

The curvature induced by assembled viral capsids initiates vesicle release, a process now mirrored in artificially constructed RNA transfer vehicles. Researchers discovered that synthetic protein assemblies, despite being smaller than traditional viral capsids, effectively induce membrane bending comparable to that of enveloped viruses, suggesting a shared biophysical principle governs vesicle formation. This finding supports the idea that even non-natural protein arrangements can initiate budding through membrane distortion, a critical step in releasing genetic material. The team fused three functional domains, membrane binding, late budding, and RNA binding, to an icosahedral protein scaffold.

These assemblies, while differing in architecture from natural capsids, demonstrated a similar capacity to initiate vesicle release, indicating a potential optimum vesicle size independent of the originating protein structure. “Although STV-C8 assemblies were generally smaller than viral capsids, they induced release of vesicles of similar size to those released by enveloped viruses,” the researchers report, highlighting the efficiency of the membrane bending mechanism. Prior work proposed that partially assembled protein protomers, as well as fully formed capsids, could induce membrane bending, and this study provides evidence supporting that hypothesis with entirely synthetic components.

The observed vesicle release, driven by these artificial assemblies, suggests a fundamental biophysical process at play, independent of the evolutionary history of viral capsid formation. This approach leverages the inherent properties of membrane curvature to facilitate RNA transfer, offering a potentially versatile platform for targeted delivery and gene editing applications, and demonstrating that the principles governing viral assembly can be successfully replicated in bottom-up designs.

Stay current

See today’s quantum computing news on Quantum Zeitgeist for the latest breakthroughs in qubits, hardware, algorithms, and industry deals.

Tags:
Avatar of Ivy Delaney

Ivy Delaney

Ivy Delaney has been working with neural networks and machine learning since the mid-nineties, back when a couple of hidden layers and a long afternoon of training counted as ambitious. She has watched the field go from academic curiosity to the thing quietly running underneath everything, and she brings that long view to quantum computing. For Quantum Zeitgeist she covers the ground where the two fields meet. That means quantum machine learning and the variational algorithms it leans on, and it also means the less glamorous but more interesting story of classical machine learning already doing real work inside quantum machines, decoding error-correcting codes, calibrating noisy hardware and learning the error models that simulators depend on. She writes about the hardware those algorithms have to run on too, and about the post-quantum cryptography scramble that the same hardware has set off. Her stories typically start with the paper, whether that is peer-reviewed work, conference proceedings or an arXiv preprint, with the source linked so you can hold a claim up against the research it came from. She is unimpressed by benchmarks that will not say what they beat, and by demonstrations that only work in the press release.

Latest Posts by Ivy Delaney: