Identification of micro-organisms by Raman spectroscopy

The fast and unambiguous identification of micro-organisms has become an important requirement not only in clinical analysis, but also in various hygiene-monitoring applications such as pharmaceutical production or food-processing technology. In all applications, it is highly desirable to minimize the micro-organism cultivation time to quickly provide patients with appropriate treatment or minimize production downtimes. However, conventional identification methods based on biochemical tests after cultivating the micro-organisms in different media have the drawback of being very time-consuming. Among the new techniques being developed for rapid micro-organism identification, Raman spectroscopy is becoming recognized as a very promising tool.

In our work, we use conventional Raman to characterize single bacterial cells as well as resonance Raman excitation in the UV-range to obtain selectively enhanced signals of DNA/RNA and aromatic amino acids (see Figure 1). Our approach is based on four steps: collection of the micro-organisms, localization of biological molecules using fluorescence-staining techniques, Raman measurements, and comparison of the Raman spectra with a database for identification purposes. We recently showed that UV-resonance Raman spectroscopy on bulk samples (up to 10⁶ cells) can lead to clear genotypic identification.¹ We have also established that at least one microcolony (approximately 10⁶ cells) is required for one measurement. Since UV excitation often leads to sample destruction, we also showed that measurements are best performed on rotated bacterial films. To obtain our first dataset, we investigated 34 strains of a bacterial organism. Combining their Raman results with a chemometric classification method, we were able to achieve an average strain recognition rate exceeding 98%.²

To achieve faster identification while avoiding the cultivation step, single-bacteria analysis is required. Micro-Raman spectroscopy with 532nm-excitation has a spatial resolution of approximately 1μm, which is quite adequate for the spectral investigation of single bacterial cells (see Figure 1A). Since this wavelength is non-resonant with any electronic excitation, the recorded spectra include the spectral signatures of the various molecules present inside the cell.³ This average molecular spectral information can then be used for phenotypic bacterial identification purposes at the single cell level without the need for a time-consuming cultivation step. We were able to achieve average strain recognition rates over 86% using a database of 50 strains.²

Figure 1. A: Micro-Raman spectra of single bacterial cells recorded with 532nm excitation. B: UV-resonance Raman spectra of bacterial bulk samples excited at 257nm.

The localization of micro-organisms in complex sample matrices such as soil or food samples or in samples such as air, in which they only occur at low concentrations, introduces another level of difficulty. This is because these samples contain many particles with sizes and shapes often similar to that of the micro-organisms. Therefore, a localization routine has to be established to differentiate biotic and abiotic particles. One possible approach is the use of fluorescent dyes. A drawback however, is that the fluorescence intensity often masks the Raman spectrum completely. One solution is to use fluorescence labels with absorption maxima occurring far from the Raman excitation wavelength. This allows for the selective localization of bacteria which can then be identified in a second step by Raman spectroscopy and chemometric methods.⁴

To improve the characterization of bacteria, the cellular compounds associated with strain and species differentiation also need to be accurately identified. For this purpose, we recently applied tip enhanced Raman spectroscopy (TERS) to the identification of single bacterial cells.⁵ TERS is a technique that combines surface enhanced Raman spectroscopy (SERS) with atomic force microscopy (AFM). In our work, we used a single silver particle located on the tip of an AFM cantilever as the SERS substrate. We were able to characterize cell surfaces with a spatial resolution of approximately 30nm and to identify the different cell wall components of S. epidermidis.

Funding of research project FKZ 13N8369 within the framework of the 'Biophotonik' program of the Federal Ministry of Education and Research, Germany (BMBF), is gratefully acknowledged.