Massachusetts Institute of Technology, United States
Introduction: Diagnostics play a critical role in medicine, yet, essential diagnostics are inaccessible to as much as 62% of the world’s population. Standard methods rely on time-consuming culturing or molecular-based tools with tedious sample preparation steps. Raman spectroscopy, however, exhibits strong potential as a point-of-care diagnostic, rapidly generating molecular fingerprints of cells via their inelastic Raman scattering of light. Here, we demonstrate its application in rapid, label-free classification of clinically relevant bacteria species, as well as antibiotic susceptibility testing in tuberculosis. For more dilute samples, we highlight two approaches: the use of 1) immunomagnetic beads to isolate and report the presence of target bacteria with their own unique Raman signature and 2) plasmonic metal nanorods to uniformly enhance the Raman signal of bacteria in liquid solutions. We also discuss ongoing work towards clinical translation including advancements in portability, automated sample and data processing.
Materials and
Methods: As one approach to Raman spectroscopy for cell identification in liquid samples, we deployed Dynabeads anti-Salmonella to bind and act as Raman reporters for Salmonella enterica, a major foodborne pathogen. Conjugates of Dynabeads anti-Salmonella and S. enterica were formed and suspended in a liquid well of deionized water (DIW) with optically transparent quartz substrate. Using the superparamagnetic property of Dynabeads, the conjugates were concentrated down to the imaging surface using a magnet immediately before Raman interrogation with a 785 nm incident laser focused at the bottom of the well (Figure 1). This magnetic concentration step semi-fixed bead-bound cells in the field of view rather than having them freely float in the liquid. A customized inverted Raman system was utilized to detect Raman spectra from liquid samples and visualize their intensities with single, 0.5 sec acquisitions at 75 mW power; polynomial fitting-assisted background-subtraction was performed using Lieberfit.
As a second approach, we synthesized plasmonic Au-nanorods and mixed them with individual suspensions of 4 species of bacteria in liquid wells of DIW (Figure 2A). Raman spectra from 30 s acquisitions were collected using the Renishaw inViaTM confocal Raman microscope with 785 nm incident laser at 30 mW power. We confirmed uniform enhancement across the liquid well by collecting the SERS spectra at signature peaks across a 100 × 200 μm² region. We also investigated the effects of cell surface charge, visualizing the interactions between nanorods and cells via cryo-electron microscopy. Background subtraction was performed using a polynomial fit with degree 70.
Results, Conclusions, and Discussions: Figure 1 summarizes the rapid detection of bacteria in liquid samples with Raman reporter Dynabeads, showing that S. enterica has a weak Raman signal on its own, while bead-containing samples have a unique, high-intensity Raman signature from the beads themselves. From reported spectra, we attribute Dynabeads’ signature peaks observed in Figure 1 (1000, 1350, and 1600 cm-1) to their polystyrene and antibody coating and note their prominence in both single and clustered samples. Conjugation to bacteria promotes bead clustering, resulting in stronger Raman signals likely due to increased polystyrene and antibody content at the spot of optical interrogation.
Alternatively, Figure 2 demonstrates the detection of bacteria in liquid samples with SERS, showing that bacteria alone have indistinguishable Raman spectra, but the addition of plasmonic nanorods yields high-intensity bacteria spectra matching reported spectra from dried samples. Little signal is generated from the nanorods themselves. Species-specific differences in intensity are likely due to differences in electrostatic interactions between nanorods and cells; tighter aggregation of positively-charged Au-nanorods around bacteria like S. epi. with more highly negative surface charge densities yields higher signal intensity than E. coli with less negative surface charge density. Uniform enhancement at signature peaks indicates monodispersity of nanoparticles in solution, compared to clumping in dried samples.
Overall, Dynabeads’ strong Raman signal highlights their ability to serve as Raman reporters, especially for samples with weak Raman signals that make it difficult to observe signature peaks for cell identification. Importantly, the intense signal observed in single conjugates demonstrates the single-cell sensitivity of this approach. The need for cell-specific antibodies, however, limits their use without a priori knowledge of target species. Contrastingly, the enhancement of Raman spectra with SERS enables the observation of signature peaks without cell-specific labels. With little signal from Au-nanorods themselves, the Raman signal originates from cells themselves and is applicable across species. Uniform enhancement enables more reliable measurements, and species-specific intensity indicates opportunity for tunable specificity via electrostatic interactions. This approach is challenged, however, at low concentrations of cells, limiting its sensitivity. Together, these two approaches cover a wide range of applications, each improving cellular diagnostics in resource-limited environments.
