J11 - Biomineralized SERS nanotags for multi-faceted disease diagnosis

Introduction: : Surface-enhanced Raman spectroscopy (SERS) nanotags offer strong potential for ultrasensitive in vitro diagnostics, but current design strategies often demand disease-specific customization, limiting scalability and clinical translation. Biomineralization, a well-established method in materials science, involves the synthesis of inorganic structures through biomolecular mediation. While this approach has been used to create nanomaterials for drug delivery and therapeutic purposes, its diagnostic potential—particularly in the context of SERS—remains largely untapped. We hypothesize that gold nanoparticles (AuNPs) synthesized via biomineralization using diverse biological materials (e.g., amino acids, proteins, exosomes, tumor cells, bacterial pathogens) will exhibit distinct optical and spectroscopic features that can be leveraged for diagnostics. We propose that these variations can be captured through SERS and analyzed using machine learning algorithms to classify the originating biomaterial. To establish proof-of-concept, we synthesized SERS nanotags using seven α-amino acids with varied physicochemical properties and characterized them using UV-Vis, DLS, TEM, and Raman spectroscopy. Then to further classify the amino acid more accurately we applied machine learning approach including PCA and LDA. This strategy demonstrates the potential of combining biomineralization and machine learning to develop a versatile, scalable diagnostic platform based on SERS nanotags.

Materials and

Methods: : AuNPs were synthesized using seven different α-amino acids, each selected for varying size, polarity, and charge. A 5 mM chloroauric acid solution was heated to boiling, then mixed with 20 mM of amino acid and 200 µL NaOH to adjust pH and initiate nanoparticle formation. Synthesized AuNPs were then characterized using UV-Vis spectroscopy, dynamic light scattering (DLS), and transmission electron microscopy (TEM) and SERS spectra. Principal component analysis (PCA) and linear discriminant analysis (LDA) were applied individually to the SERS spectra and in combination to analyze spectral patterns and classify the biomineralized AuNPs.

Results, Conclusions, and Discussions:: Results and

Discussion: UV-Vis spectroscopy showed absorbance peaks between 530–580 nm, confirming nanoparticle formation (figure 2-A). DLS measurements indicated particle sizes between 16–50 nm with consistent Zeta potential and low polydispersity (figure 2- B, C). TEM confirmed the spherical morphology and biomolecule coatings of the nanoparticles (figure 2- D). SERS spectra revealed unique Raman fingerprints for each amino acid (Figure E), and signal intensity increased with higher reporter dye concentrations (Figure 2- F). Machine learning approaches effectively classified the nanotags. PCA and LDA each distinguished spectral clusters based on amino acid identity, with the PCA-LDA combination providing the clearest separation (Figures 2- G, H, I).

Conclusion: We successfully synthesized biomineralized SERS nanotags from seven amino acids and used machine learning to classify them based on Raman data. This work demonstrates the feasibility of integrating biomineralization and statistical learning to support diagnostic development. Our next step is to expand this platform to classify disease-relevant proteins and distinguish healthy cells from cancerous cells across different tissue types. Ultimately, we aim to classify healthy versus diseased cells using patient-derived exosomes, laying the foundation for a modular, patient-specific cancer diagnostic platform based on biomineralized SERS nanotags.

Acknowledgements and/or References (Optional)::