Surface-enhanced Raman Spectroscopy Of Serum Protein Based On Membrane Electrophoresis

The research for Life Science has gone to post-genome era which focuses on the research of proteomics. In this study, the combination of surface-enhanced Raman scattering (SERS) and membrane electrophoresis (ME) was applied in the biochemical analysis of serum proteins. Multivariate statistical analysis was used to construct diagnostic algorithms to identify ABO blood type and detect nasopharyngeal, colorectal, gastric, and liver cancers. The major results are as follows:(1) Identify blood type based on one-way ME-SERS. This method was applied to analyze blood types of118healthy samples. Comapring the average SERS spectra and PCA loading plots of serum globulin from different blood types, we find that globulin from different blood types have similar spectral structures, but their abilities to approach silver surfaces seem to differ. Based on the combination of principle component analysis and linear discriminant analysis (PCA-LDA) model and receiver operating characteristic (ROC) analysis, the best-averaged sensitivities and specificities for each dataset were (blood types A and B:90.0%,95.0%; blood types A and O:80.0%,83.9%; Blood types B and O:95.0%,90.3%; blood types AB and A:97.3%,96.7%; blood types AB and B:94.6%,95.5%; blood types AB and O100%,100%). The corresponding area under ROC curves (AUCs) were0.978,0.847,0.965,0.978,0.986, and1.000, respectively. These results suggest that one-way ME-SERS method can study blood types at molecular level, showing great potential in providing researchers with novel insights.(2) Detect hepatocellular carcinoma (HCC) by the combination of one-way ME-SERS and drop coating deposition (DCD). The introduction of DCD will further improve the intensity and reproducity of serum protein signal. PCA-LDA analysis of the SERS spectra of serum proteins reveals that the HCC group (n=48, including the AFP-negative results and AFP-positive results) could be unambiguously discriminated from the normal group (n=30) with100%diagnostic sensitivity and specificity, avoiding the false-negative results from AFP-negative results.50SERS spectra of albumin (normal20, HCC30) and50SERS spectra of globulin (normal20, HCC30) were employed to develop partial least squares (PLS) prediting models, respectively. Both of albumin and globulin PLS models successfully predicted the unidentified subjects (normal10, HCC18) with diagnostic accuracies of100%, showing great potential in providing a fast and invasive method for HCC detection.(3) Detect colorectal cancer by one-way ME-SERS. Glacial acetic acid was introduced as the aggreagating agent in order to provide additional SERS enhancement without the generation of signal interference. The membrane itself was also used as the control to improve the accuracy and efficiency of protein separation. The optimized method combined with PCA-LDA analysis achieve diagnostic accuracies of100%and99.5%based on the albumin and globulin average SERS spectra (normal103, colorectal cancer103). Additionally, the SERS spectra of160albumin samples (normal80, colorectal cancer80) and160globulin samples (normal80, colorectal cancer80) were used to construct PLS predicting models, respectively, both of which successfully predicted the unidentified subjects (normal23, colorectal cancer23) with diagnostic accuracies of93.5%, showing great potential in offering a fast method for colorectal cancer detection.(4) Establish forward and reverse ME-SERS to detect the nasopharyngeal cancer group (n=128), the colorectal cancer (n=109), the gastric cancer group (n=133), and the liver cancer group (n=133). Compared to one-way ME-SERS, forward and reverse ME-SERS is not only faster and more efficient, but also can differentiate four cancer types from each other. Diagnostic sensitivities of95.3,93.2,90.8, and83.3per cent, and specificities of98.2,98.1,93.2and98.1per cent, respectively, were achieved based on PCA-LDA analysis. The correspongding AUC values were0.996,0.991,0.973, and0.987, respectively. The results indicate that this approach has great potential to identify different types of cancer.