Surface-enhanced Raman spectroscopy introduced into the International Standard Organization (ISO) regulations as an alternative method for detection and identification of pathogens in the food industry

We show that surface-enhanced Raman spectroscopy (SERS) coupled with principal component analysis (PCA) can serve as a fast, reliable, and easy method for detection and identification of food-borne bacteria, namely Salmonella spp., Listeria monocytogenes, and Cronobacter spp., in different types of food matrices (salmon, eggs, powdered infant formula milk, mixed herbs, respectively). The main aim of this work was to introduce the SERS technique into three ISO (6579:2002; 11290–1:1996/A1:2004; 22964:2006) standard procedures required for detection of these bacteria in food. Our study demonstrates that the SERS technique is effective in distinguishing very closely related bacteria within a genus grown on solid and liquid media. The advantages of the proposed ISO-SERS method for bacteria identification include simplicity and reduced time of analysis, from almost 144 h required by standard methods to 48 h for the SERS-based approach. Additionally, PCA allows one to perform statistical classification of studied bacteria and to identify the spectrum of an unknown sample. Calculated first and second principal components (PC-1, PC-2) account for 96, 98, and 90% of total variance in the spectra and enable one to identify the Salmonella spp., L. monocytogenes, and Cronobacter spp., respectively. Moreover, the presented study demonstrates the excellent possibility for simultaneous detection of analyzed food-borne bacteria in one sample test (98% of PC-1 and PC-2) with a goal of splitting the data set into three separated clusters corresponding to the three studied bacteria species. The studies described in this paper suggest that SERS represents an alternative to standard microorganism diagnostic procedures. Graphical Abstract New approach of the SERS strategy for detection and identification of food-borne bacteria, namely S. enterica, L. monocytogenes, and C. sakazakii in selected food matrices Electronic supplementary material The online version of this article (doi:10.1007/s00216-016-0090-z) contains supplementary material, which is available to authorized users.


Characteristics of Salmonella spp., L. monocytogenes, and C. sakazakii
Salmonella enterica common bacteria found in rotten or unwashed food is one of the most important foodborne pathogens worldwide and the second most frequently reported zoonotic agent in the European Union (EU) after thermotolerant Campylobacter. In 2014, a total of 88,238 confirmed salmonellosis cases were reported by 27 member states (MS) of the European Union, resulting in a notification rate of 23.4 cases per 100,000 population [28].
Most people infected with Salmonella develop diarrhea, fever, and abdominal cramps. The Salmonella infection may spread from the intestines to the blood stream, and then to other body sites and may even cause death [29]. Therefore the fast and simple detection of Salmonella in food is needed.
Another serious infection is listeriosis usually caused by eating food contaminated with L.
monocytogenes. The disease primarily affects newborns, pregnant women, older people, and adults with weakened immune systems. A person with listeriosis usually has invasive infection. Pregnant women may experience fever and other non-specific symptoms, such as fatigue and aches, followed by fetal loss or bacteremia and meningitis in their newborns. Cronobacter sakazakii, formerly Enterobacter sakazakii (according to an old system of nomenclature), is a germ that can live in very dry places. It has been found in dry foods, like powdered baby formula, powdered milk, herbal teas, herbs, and starches. In babies, C.
sakazakii germs usually get in the blood or make the lining of the brain and spine swell (meningitis). Up to 4 out of 10 babies with meningitis from C. sakazakii can die. Infections caused by Cronobacter spp. are also dangerous for older people and people whose bodies have trouble fighting germs, like people with HIV, organ transplants, or cancer [33]. The fast detection of C. sakazakii, especially in powdered milk and powdered infant formula, is very important.

Validation of the PCA model
For PCA calculation the Unscrambler@ software were used. In that software we asked for cross-validation of obtained results by uncertainty test (using the optimal number of PC).
It is common practice to use cross-validation for determining the number of components and then use that number in further modeling. In our case the calculated scores have been used for building a classification model using principal component regression. The built model performs very well with an R-squared of 0.99 and correlation coefficient of 0.95 (see Fig. S6, Supplementary Materials) for adjusted P-value (conservatively) to 0.05. Our calculations proved that the validated data (cross-validation data) and calibrated (for p-value) are in the excellent agreement (see Fig. S6). Moreover, the predicted and the reference values of PC components are also highly correlated (see Fig. S6). Therefore the proposed model is rational for the classification purposed of our studied data.

Sensitivity and specificity of used method
According the PCA calculation all obtained PC scores were nicely clustered with large distance among clusters (S. Typhimurium, L. monocytogenes and C. sakazakii , e.g., see Based on the above equations and the gathered data (with no wrong clustering) the calculated both sensitivity and specificity give 100%. Taking into account that all data in the manuscript are very limited in comparison to the real condition, we are far long from such rigorous statement, thus we decided not to introduce that values to the manuscript.

Reproducibility of bacterial SERS signals
The reproducibility of recorded SERS signals plays a crucial role in the analytical and biomedical applications of SERS technique. The average standard deviation (Av. STD) of the SERS signals three bacteria Salmonella Typhimurium, L. monocytogenes and C. sakazakii were calculated and presented in the Table S1.  Fig. S7 SERS spectra of C. sakazakii, recorded from different spots within the same sample. The excitation wavelength was at 785 nm, laser power was 5 mW, and acquisition time was 60 s