Presentation Information
[P04-560]Machine learning-assisted screening of blue-colored triangular gold nanoplate-forming peptides
○Shogo Saito1, Masayoshi Tanaka1, Mina Okochi1 (1. Institute of Science Tokyo (Japan))
Keywords:
peptide,biomineralization,Gold nanoparticle,machine learning
[Purpose]
Gold nanoparticles (AuNPs) exhibit morphology-dependent optical properties. In particular, triangular gold nanoplates (AuNPLs) display strong plasmonic absorption in the visible–near infrared region and often appear blue due to their anisotropic structure. Such optical characteristics are attractive for applications in biosensing and photothermal therapy. Our research group previously developed a peptide-mediated one-pot synthesis method that enables shape-controlled AuNP formation from tetrachloroauric acid at room temperature. However, only a few peptides have been used to synthesize blue-colored AuNPLs, and improving synthesis efficiency remains difficult. In this study, we aimed to design peptides capable of producing blue-colored anisotropic AuNPs by combining peptide array screening with machine learning.
[Method]
A peptide library was constructed based on 200 sequences obtained in our previous study (M. Tanaka et al., Nanoscale Adv., 2019, 1, 71-75). Peptide arrays were synthesized and evaluated for their ability to mineralize AuNPs and generate blue-colored particles. Machine learning models were then built using a random forest algorithm. Two classification models were prepared: one to predict mineralization activity and another to predict AuNPL formation. Based on model predictions, candidate peptides were selected, and the next peptide array was synthesized. This cycle of prediction, peptide array synthesis, and evaluation was repeated twice to refine peptide candidates.
[Results]
Using this iterative screening approach, four peptide sequences capable of forming blue-colored AuNPs were identified. Among them, the BR2-3 peptide (RWGGGIGGWN) produced uniform triangular nanoplates with a synthesis efficiency of 33%. More than 95% of the generated particles were triangular or decahedral in shape, with diameters ranging from 10 to 100 nm. The colloidal solutions exhibited a blue coloration, consistent with the formation of anisotropic nanoplates. Cellular uptake of AuNPs synthesized by the four peptides was evaluated using MDA-MB-231 cells. AuNPs synthesized using the BR2-3 peptide showed higher cellular uptake compared with particles generated by the other peptides.
[Consideration]
The integration of peptide array screening and machine learning enabled efficient identification of peptides that generate blue-colored anisotropic AuNPs. Furthermore, the particle size range obtained in this study is suitable for cellular uptake, indicating potential applicability in biomedical fields. These results demonstrate that data-driven peptide screening can accelerate the discovery of morphology-controlling mineralization peptides.
[Conclusion]
Machine learning-assisted peptide screening identified peptides that selectively generate blue-colored triangular gold nanoplates with efficient cellular uptake.
Gold nanoparticles (AuNPs) exhibit morphology-dependent optical properties. In particular, triangular gold nanoplates (AuNPLs) display strong plasmonic absorption in the visible–near infrared region and often appear blue due to their anisotropic structure. Such optical characteristics are attractive for applications in biosensing and photothermal therapy. Our research group previously developed a peptide-mediated one-pot synthesis method that enables shape-controlled AuNP formation from tetrachloroauric acid at room temperature. However, only a few peptides have been used to synthesize blue-colored AuNPLs, and improving synthesis efficiency remains difficult. In this study, we aimed to design peptides capable of producing blue-colored anisotropic AuNPs by combining peptide array screening with machine learning.
[Method]
A peptide library was constructed based on 200 sequences obtained in our previous study (M. Tanaka et al., Nanoscale Adv., 2019, 1, 71-75). Peptide arrays were synthesized and evaluated for their ability to mineralize AuNPs and generate blue-colored particles. Machine learning models were then built using a random forest algorithm. Two classification models were prepared: one to predict mineralization activity and another to predict AuNPL formation. Based on model predictions, candidate peptides were selected, and the next peptide array was synthesized. This cycle of prediction, peptide array synthesis, and evaluation was repeated twice to refine peptide candidates.
[Results]
Using this iterative screening approach, four peptide sequences capable of forming blue-colored AuNPs were identified. Among them, the BR2-3 peptide (RWGGGIGGWN) produced uniform triangular nanoplates with a synthesis efficiency of 33%. More than 95% of the generated particles were triangular or decahedral in shape, with diameters ranging from 10 to 100 nm. The colloidal solutions exhibited a blue coloration, consistent with the formation of anisotropic nanoplates. Cellular uptake of AuNPs synthesized by the four peptides was evaluated using MDA-MB-231 cells. AuNPs synthesized using the BR2-3 peptide showed higher cellular uptake compared with particles generated by the other peptides.
[Consideration]
The integration of peptide array screening and machine learning enabled efficient identification of peptides that generate blue-colored anisotropic AuNPs. Furthermore, the particle size range obtained in this study is suitable for cellular uptake, indicating potential applicability in biomedical fields. These results demonstrate that data-driven peptide screening can accelerate the discovery of morphology-controlling mineralization peptides.
[Conclusion]
Machine learning-assisted peptide screening identified peptides that selectively generate blue-colored triangular gold nanoplates with efficient cellular uptake.
Comment
To browse or post comments, you must log in.Log in