A dual-biomarker-controlled DNA logic platform assisted with multivariate statistical analysis for accurate and high-throughput recognition of cancer cells

Abstract

Present DNA logic platforms mainly focused on distinguishing a specific kind of cancer cell from several other types of cell lines, while limited efforts have been paid to high-throughput recognition of cancer cells. A new DNA logic platform coupled with multivariate statistical analysis was developed to classify seven kinds of cancer cells. Four confined DNA circuits containing localized catalytic hairpin assembly (LCHA) and apurinic/apyrimidinic endonuclease 1 (APE1)-assisted circuits were conjugated to a protein of streptavidin (SA) for DNA logic platform. The mismatch minimizes the circuit leakage of the confined DNA circuits by inhibiting the incorrectly intramolecular structure switch. Benefitting from the spatial-confined effect, the DNA logic platform as a modified AND logic gate sensitively responded to the dual intracellular biomarkers, thereby generating strong dual fluorescent output signals in living cells. Based on the fluorescent cell imaging, multidimensional data collected from 50 individual cells were concatenated as a dataset for multivariate statistical analysis. Principal component analysis (PCA) shows distinct clustering among the seven kinds of cancer cell lines. A random forest algorithm obtained the specificity and sensitivity of 100%. Therefore, the DNA logic platform coupling with multivariate statistical analysis shows promising potential for cancer diagnosis and treatment.