The onset and progression of neurodegenerative diseases are closely linked to protein misfolding and pathological aggregation. Among these, abnormal aggregation of TDP-43 (trans-activator response DNA-binding protein of 43 kDa) constitutes a core pathological hallmark in conditions such as amyotrophic lateral sclerosis (ALS) and frontotemporal dementia (FTLD). — with nuclear depletion and cytoplasmic fibrillar aggregates of TDP-43 detectable in approximately 97% of ALS patients and 50% of FTLD patients. These aggregates not only disrupt RNA metabolic homeostasis but also directly mediate neuronal toxicity and disease progression. However, due to the extreme conformational variability of TDP-43's low-complexity domain (LCD) within its intrinsically disordered region (IDR), targeted intervention at its aggregation core has long faced technical barriers. To date, no effective small-molecule or biomolecular therapies have been approved.

Recently, a collaborative team led by Professor Dan Li from the Centre for Ultrafast Science at the Zhangjiang Advanced Research Institute, Bio-X Institute, Shanghai Jiao Tong University, alongside researchers from the School of Life Sciences and Technology at ShanghaiTech University and the Shanghai Institute of Organic Chemistry, Chinese Academy of Sciences, published a study titled ‘De novo design of protein binders to stabilise monomeric TDP-43 and inhibit its pathological aggregation’ in PNAS. This study pioneers the use of artificial intelligence-driven de novo protein design to develop novel binders that specifically and highly affinity-binding to the LCD aggregation core of TDP-43. These binders effectively inhibit pathological TDP-43 aggregation in both in vitro and cellular models, offering a novel therapeutic strategy for neurodegenerative diseases.

The physiological function of TDP-43 relies on its structured N-terminal domain (NTD) and RNA recognition motifs (RRM1/2), whereas the C-terminal LCD serves as the primary driver of pathological aggregation. Within the LCD, residues 319–335 adopt an α-helical conformation under physiological conditions. In disease states, a ‘α-helix → β-sheet’ conformational shift occurs, leading to the assembly of polymorphic fibrillar aggregates. With the core objective of ‘stabilising the α-helical conformation of the TDP-43 CR domain,’ the research team established a comprehensive workflow spanning computational design to functional validation. The ligand design phase commenced by generating candidate scaffolds via diffusion modelling. Sequences were then optimised using deep learning tools to enhance interactions with key sites within the CR domain. Energy screening and structural prediction ultimately identified high-potential ligand candidates (Figure 1).

Figure 1. Workflow diagram of this study

During the in vitro validation phase, the research team employed Tht assays specifically detecting β-folded aggregates alongside electron microscopy observations. This revealed that certain binding agents effectively inhibited TDP-43 aggregation, with the inhibitory effect exhibiting a positive correlation with binding affinity (Figure 2).

Figure 2. Inhibition of TDP-43 fibrillar aggregation by a de novo designed binding protein

To elucidate the mechanism of action, structural prediction revealed that this highly efficient binder forms tight hydrophobic interactions with the TDP-43 CR helix in a specific conformation. Subsequent NMR experiments further confirmed its specific binding to the CR helix region.

Subsequently, site-directed mutagenesis experiments established that this hydrophobic interaction constitutes the core mechanism by which the ligand ‘locks’ the α-helical conformation and blocks aggregation. Mutating either key residues in the CR region of TDP-43 or interface residues in the ligand resulted in loss of binding and inhibitory activity.

Finally, at the cellular level, the team validated this binding agent in two classical cell models, demonstrating its ability to significantly reduce the ratio of TDP-43 aggregation within the nucleus versus the cytoplasm, while also decreasing levels of disease-associated biomarkers.

In summary, this study provides the first validation of the feasibility of inhibiting pathological aggregation by stabilising the α-helical conformation of specific motifs within the intrinsically disordered region of TDP-43, offering novel insights for precisely targeting core pathogenic factors in neurodegenerative diseases. The researchers noted that this strategy is not only applicable to TDP-43 but may also be extended to other intrinsically disordered proteins involved in amyloid aggregation in neurodegenerative diseases, such as pathogenic proteins associated with Alzheimer's and Parkinson's diseases. Concurrently, the team emphasised that small-molecule protein therapeutics still face numerous challenges in clinical application, including blood-brain barrier penetration, in vivo stability, and immunogenicity issues, all of which require further resolution in the future.

This collaborative study involved the School of Life Science and Technology at ShanghaiTech University, the Bio-X Institute at Shanghai Jiao Tong University, the Zhangjiang Advanced Research Institute, and the Shanghai Institute of Organic Chemistry, Chinese Academy of Sciences. Corresponding authors include Professor Xu Wenqing and Associate Researcher Wang Zhizhen from ShanghaiTech University's School of Life Science and Technology; Professor Li Dan from Shanghai Jiao Tong University, Bio-X Institute, and Zhangjiang Advanced Research Institute; and Researcher Liu Cong from the Shanghai Institute of Organic Chemistry, Chinese Academy of Sciences. Sun Gangyu, PhD candidate at the School of Life Science and Technology, ShanghaiTech University, and Li Xiang, PhD candidate at Shanghai Jiao Tong University, are the co-first authors of this paper.

Paper link: https://www.pnas.org/doi/10.1073/pnas.2505320122