Surface-enhanced Raman scattering (SERS)–based immunosystem for ultrasensitive detection of the 90K biomarker

The research and the individuation of tumour markers in biological fluids are currently one of the main tools to support diagnosis, prognosis, and monitoring of the therapeutic response in oncology. Although the identification of tumour markers in asymptomatic patients is crucial for early diagnosis, its application is still limited by the relatively low sensitivity and the complexity of existing methods (i.e. ELISA, mass spectrometry). We developed an easy, fast, and ultrasensitive surface-enhanced Raman scattering (SERS)–based system, for the detection and quantitation of the LGALS3BP (90K) biomarker that was used as a model, based on the development of antibody-functionalized nanostructured gold surfaces. The detection system was effective for the ultrasensitive detection and characterization of samples of different biochemical compositions. In conclusion, this work could provide the foundation for the development of a medical diagnostic device with the highest predictive power when compared with the methods currently used in cancer diagnostics. Electronic supplementary material The online version of this article (10.1007/s00216-020-02903-2) contains supplementary material, which is available to authorized users.


Fig. S1
Typical SERS spectrum with highlighted some unwanted features The raw SERS-Raman spectra are affected by unwanted features and components that prevent the analysis of the meaningful part of the spectrum, so typically a spectra processing is needed.
In the case of the spectra acquired for this manuscript the raw signal typically appears as reported in the figure S1 (this is an example of a clean SERS substrate).
The unwanted components to remove are 1) the residual Rayleigh scattering; 2) random spikes; 3) the background fluorescence that is often present in a Raman spectrum. Then, for the sake of uniformity, the residual signal is normalized in the interval [0;1] ready to be processed with statistical methods (e.g. Principal Components Analysis).

1)
The residual Rayleigh signal is removed by cropping the spectrum from 350 cm-1 as reported in the manuscript.

2)
The spikes are removed by applying to the residual signal the MATLAB™ spikes removal function "medfilt1" [1] that replaces every point of a signal by the median of that point and a specified number (5, in the case of the present spectra) of neighbouring points.

3)
The background fluorescence is removed by applying a recursive procedure similar to that reported in Zhao et al. [2] and here briefly summarized: an iterative smoothing procedure is applied to the residual spectrum by using the MATLAB™ smoothing function "smooth" [3] and, after each iteration the raw signal and the smoothed signal are compared pixel by pixel to create a new partially smoothed spectrum. In this new spectrum the minima are retrieved from the raw signal and the rest of the signal is retrieved from the smoothed spectrum. Then, the procedure is applied again. After some iterations the result is a heavily smoothed spectrum anchored to the minima of the original spectrum with the maxima (the SERS peaks) completely levelled, ready to be subtracted to the raw signal to obtain a SERS spectrum without fluorescence background. The final result of this procedure is shown in the next figure S2, where in the top graph the red line represents the raw cropped data, the blue line represents the result of the iterative procedure and the green line represents the raw data subtracted with the background fluorescence (i.e. green data = red datablue data). 4) Finally, the background-subtracted data are normalized in the interval (0;1) to be uniformed for any statistical procedure, as shown in the bottom graph of figure S2.  after the first step of functionalisation a spectral feature can be observed at 612 cm -1 that was assigned to C-S stretching of lipoic acid [4,5]. d) spectrum of linker-MatoS after normalization and background subtraction ; two spectral features at 612 cm -1 and 642 cm -1 ; f) 1959Cr-MatoS spectrum after normalization and background subtraction; g) representative SERS spectrum in presence of the 90K antigen (90K-1959Cr-MatoS that shows the presence of the spectral features at 821 cm -1 and 1521-1604 cm -1 . The features at 612-640 cm -1 are ascribed to the previous functionalisation process. h) 90K-1959Cr-MatoS spectrum after normalization and background subtraction. These spectral features were assigned to C-N stretching, C-C stretching and N-H bending (amide II band; 1521 cm -1 ) and to the contribution of amino acid side chain residues such as tyrosine (642 and 821 cm -1 ) and phenylalanine (1604 cm -1 ) [6,7]