Plasmonic Molecular Entrapment for Label-Free Methylated DNA Detection and Machine-Learning Assisted Quantification

Abstract

Epigenetic DNA methylations are linked to the activation of oncogenes and inactivation of tumor suppressor genes. A reliable and label-free method to quantitatively measure DNA methylation levels is essential for diagnosing and monitoring methylation-related diseases. Herein, plasmonic molecular entrapment (PME) method assisted SERS as facile strategy for trapping and label-free sensing of DNA methylation, utilizing in situ surface growth of plasmonic particle in the presence of target analytes, are developed. This highly sensitive and adaptable technique forms hotspot sites around target analytes, overcoming mismatch geometrical properties and producing a strong electromagnetic field that leads to significant SERS signal enhancement. The PME method effectively profiles and quantifies DNA methylation, demonstrating robust capabilities for DNA analysis. A logistic regression (LR)-based machine learning accurately quantifies and classifies methylation levels in clinical serum samples of colorectal cancer and normal patients with high sensitivity, specificity, and accuracy, highlighting the feasibility of this technique. The developed PME method combined with machine learning offers promising sensing techniques for disease screening and diagnosis, marking a significant advancement in disease detection and patient care.

Keywords: DNA methylation; hotspot engineering; label‐free diagnosis; plasmonic materials; surface‐enhanced Raman scattering.

Figure 1

Schematic illustration of the PME method for label‐free DNA methylation detection and quantification.

Figure 2

Development of PME method. a) Schematic illustration of the PME method process, SEM images of AuS after PME b) without target analyte and c) with target analyte, d) HR‐TEM and TEM (inset) images of the laminated area on AuS surface in the presence of target analyte, e) SERS enhancement in PME development (scale bar refers to SERS intensity), f) Real‐time monitoring of SERS signal changes at 1 071 cm⁻¹ during PME process, g) SERS performance of PME method at various 4‐ATP concentration (scale bar refers to SERS intensity).

Figure 3

Material characterization of PME method in the presence of target DNA. a) SEM image of AuS after PME, b) TEM (left) and high resolution‐TEM (right) images of the laminated area on AuS surface, c) elemental mapping on the laminated area, d) electromagnetic field distribution of PME, e) UV–vis absorbance spectra and f) SERS enhancement of PME method for DNA detection (scale bar refers to SERS intensity), g) Raman signal enhancement mechanism of PME method.

Figure 4

SERS performance of PME method for DNA profiling. a) SERS performance of the PME method at various DNA concentrations, b) standard curves for quantitative analysis of DNA, c) signal uniformity test with 100 different points, d) reproducibility test with ten independent tests, e) SERS spectra (scale bar refers to SERS intensity) and f) signal intensity analysis of various DNA length.

Figure 5

The PME method for DNA methylation quantification analysis. a) SERS spectra of unmethylated and methylated DNA (scale bar refers to normalized SERS intensity), b) SERS spectral markers for DNA methylation, and c) SERS intensity comparison of blank (n = 10), unmethylated (n = 10), and methylated DNA (n = 10) at methylation marker peaks using one‐way ANOVA followed by Tukey's post‐hoc test (*p < 0.05, **p < 0.01, ***p < 0.001, and ****p < 0.0001). Methylation degree analysis d) SERS spectra and magnification band at 1 430 to 1 530 cm⁻¹ (inset) (scale bar refers to normalized SERS intensity), e) Raman intensity comparison performed using one‐way ANOVA followed by Tukey's post‐hoc test (n = 10 per groups), and f) concentration dependent analysis at methylation marker of 1 482 cm⁻¹. DNA size analysis (error bar refers to standard deviation), g) SERS spectra and magnification band at 1 430–1 530 cm⁻¹ (inset) (scale bar refers to normalized SERS intensity), h) Raman intensity comparison performed using one‐way ANOVA followed by Tukey's post‐hoc test (n = 10 per groups), and i) concentration dependent analysis at methylation marker (error bar refers to standard deviation).

Figure 6

DNA methylation quantification and classification in human serum. a) schematic illustration of LR‐assisted machine learning model, b) SERS spectra of human serum and DNA in serum (scale bar refers to normalized SERS intensity), c) selected Raman peak and intensity comparison of human serum (n = 100) and DNA (n = 100) using t‐test (*p < 0.05, **p < 0.01, ***p < 0.001, and ****p < 0.0001), d) ROC curve and confusion matrix (inset) of human serum and DNA, e) SERS spectra of unmethylated and methylated DNA in human serum (scale bar refers to normalized SERS intensity), f) Raman intensity comparison using t‐test and magnification of selected Raman peak (inset) of unmethylated (n = 100) and methylated (n = 100) DNA, g) ROC curve and confusion matrix (inset) of unmethylated and methylated DNA, h) SERS spectra of various methylation levels and magnification of selected Raman peak (inset) (scale bar refers to normalized SERS intensity), i) quantification of global DNA methylation levels in human serum, j) ROC curve and confusion matrix (inset) of various methylation levels.

Figure 7

Clinical application. a) SERS spectra and intensity comparison at 1482 cm⁻¹ (inset) of clinical serum from normal (n = 200) and CRC (n = 400) samples using t‐test (*p < 0.05, **p < 0.01, ***p < 0.001, and ****p < 0.0001), b) methylation prediction comparison of normal (n = 200) and CRC (n = 400) samples using t‐test, c) ROC curve and confusion matrix (inset) of normal and CRC patients, d) SERS spectra and intensity comparison at 1482 cm⁻¹ (inset) of CRC stages using one‐way ANOVA followed by Tukey's post‐hoc test (n = 100 per groups), e) methylation prediction comparison of CRC stages using one‐way ANOVA followed by Tukey's post‐hoc test (n = 100 per groups), and f) ROC curve and confusion matrix (inset) of CRC stages patients.