Info
Background:
Alzheimer’s disease (AD) and cerebral amyloid angiopathy (CAA) are characterized by the pathological deposition of amyloid-β (Aβ) peptides, which exist in different isoforms with distinct levels of neurotoxicity, including the most studied Aβ40 and Aβ421. Moreover, parenchymal plaques in AD are enriched in Aβ42, whereas vascular deposits in CAA are predominantly Aβ402. Distinguishing parenchymal and vascular Aβ deposition is important for defining AD subtypes, creating a need for isoform-selective probes. However, current imaging modalities detect total amyloid burden but lack the resolution required for mechanistic and diagnostic applications3,4.

Hypothesis/Objective:
Although aggregation and toxicity have been widely studied, the structural basis of isoform-specific roles remains unclear5,6. Recent advances in high-resolution cryo-electron microscopy (cryo-EM) structures have provided new opportunities for rational probe design7,8. This study aims to develop Aβ40-selective probes based on cryo-EM structural insights for molecular imaging to differentiate Aβ deposits in subtypes of AD.

Methods:
Cryo-EM structures of Aβ40 and Aβ42 fibrils were first analyzed to identify isoform-specific topological features and potential ligand-binding pockets. Based on the analysis, structure-guided design was performed by integrating known fluorescent probe scaffolds with differential hydrophobic cavities from the Cryo-EM structures. Moreover, molecular docking and AI-assisted modeling were used to predict binding modes and affinities. The designed probes were synthesized and evaluated in vitro against Aβ monomers, oligomers, and fibrils to determine binding affinity and selectivity toward Aβ40 over Aβ42. Non-specific binding to tau and α-synuclein aggregates was also assessed. Furthermore, brain section staining and co-localization with CRANAD-3 (a well-established Aβ fluorescence probe) and Aβ40/Aβ42-specific antibodies were performed to evaluate specificity. In vivo two-photon imaging and PET imaging were conducted in AD mouse models to assess vascular amyloid labeling and in vivo performance.

Results:
Structural analysis revealed that Aβ42 fibrils adopt a compact architecture with narrow hydrophobic channels, whereas Aβ40 fibrils exhibit a more open conformation with larger hydrophobic cavities capable of accommodating flexible ligands. Based on this structural distinction, CRANAD-191 was designed to match the unique cavity of Aβ40 fibrils. As expected, CRANAD-191 showed high affinity for Aβ40 aggregates (Kd = 61.58 nM), with significantly lower affinity for Aβ42 aggregates and oligomers (2.6 μM and 1.9 μM, respectively), corresponding to > 30-fold selectivity for Aβ40. Notably, binding to other Aβ species, tau, and α-synuclein aggregates was negligible (Kd > 10 μM). Docking studies further supported this selectivity, showing that CRANAD-191 binds within a large hydrophobic cavity in Aβ40, while only surface interactions were observed for Aβ42. Brain section staining showed strong labeling of amyloid deposits with high co-localization with CRANAD-3 and Aβ40 antibodies, but weak overlap with Aβ42 antibodies. In vivo two-photon imaging further confirmed robust vascular amyloid labeling in AD mice.

For PET imaging, CRANAD-191 was modified to obtain CRANAD-280, which retained high affinity for Aβ40 aggregates (Kd = 237 nM) and >30-fold selectivity, with Kd values for other Aβ species and misfolded proteins all >10 μM. Then, 18F-CRANAD-280 was successfully synthesized via nucleophilic fluorination of a tosyl precursor with RCY > 95%. PET imaging demonstrated that the probe has a suitable PK/PD profile for brain imaging, as the signal intensity ratio of 2 min to 30 min was 2.60. As expected, PET imaging results showed significantly higher brain uptake in AD mice compared to wild-type controls (5.51 vs 4.19 %ID/cc).

Conclusion:
This study demonstrates that cryo-EM-guided design enables the development of highly Aβ40-selective imaging probes. The probes exhibit strong in vitro and in vivo performance and potentially allow differentiation between vascular and parenchymal amyloid deposition. These findings provide new tools for investigating isoform-specific roles of Aβ and offer potential strategies for improved diagnosis and disease stratification in AD subtypes.

Image/Figure:
Click to view full size
Click to view full size

Image/Figure Caption:
Cryo-EM structure analysis revealed a larger hydrophobic cavity in Aβ40 fibrils compared to Aβ42, enabling selective probe development.
The fluorescent probe CRANAD-191 showed high affinity and selectivity for Aβ40 aggregates, which was confirmed by brain staining and antibody co-localization.
In vivo two-photon imaging demonstrated selective labeling of vascular amyloid.
The optimized PET probe exhibited high brain uptake (>2 %ID/cc) and enabled differentiation of AD and WT mice.
These results highlight the potential of structure-guided design for isoform-selective amyloid imaging.

Author
Jiang Yu
Massachusetts General Hospital and Harvard Medical School