Our ultimate goal in this research is to establish a Raman-activated cell sorting (RACS) platform capable of analysing diverse sample types, to match the wide applicability of fluorescence-activated cell sorting (FACS) systems. FACS, relying on the detection of fluorescence, is a widely used platform where microbes of interest that have been marked using specific fluorescent stains according to their metabolic functions or genetic identity can be counted and sorted. It achieves very high throughput, enabling measurements and sorting of up to 100,000 cells per second.
Our platform evolved by inspiration from FACS. It is based on our realisation that Raman microspectroscopy could be used to identify and thus sort individual cells that perform a specific function within a complex microbial community. Raman technology has great potential in aquatic and environmental microbiology as it can measure the molecular composition of microbes (e.g., lipids, proteins, carbohydrates, DNA/RNA, carotenoids, or introduced markers) in situ at the level of a single cell. Raman measurements of individual microbes can thus be used to detect cells possessing biomolecules of interest (e.g., carotenoids, cytochromes) or cells labelled with a stable isotope (e.g., deuterium, 13C). This yields enormous versatility, allowing sorting of cells on the basis of extremely diverse ecological functions without the need for specific fluorescent stains.
Our collaborators at the University of Vienna (Prof. Michael Wagner, Prof. David Berry, Dr Marton Palatinszky, and Dr Fatima C. Pereira) had developed a manual sorting method to collect taxa of interest from within a community. However, manual sorting was extremely laborious because it involved fishing for cells of interest in a glass capillary (based on trapping of individual cells using optical tweezers and their Raman measurement) followed by very slow and careful transport into a cell-free region for the recovery of sorted cells (see supplementary Fig. 1 in our paper for detailed procedures). Our collaboration had the goal of increasing the low throughput of this manual sorting (in practice, 1–2 cells per hour).
The key to increasing throughput was to accelerate and automate the handling of the cells. Microfluidics provided an approach to manipulate the samples. We engineered a platform built around a microfluidic device in which flow carries a sample of cells from a community and we integrated optical tweezers to trap single cells as they pass. Measurements of the Raman spectrum within the tweezers using a commercial system allow our automated sorting software to determine when a cell has been captured. It then performs a more detailed Raman interrogation to determine whether the cell should be sorted or discarded depending on whether the cell has the characteristic of interest (i.e., the chemical moiety of interest or isotopic labelling). Initially, in place of the optical tweezers, we had considered the use of a single-file cell stream with measurement of cells as they passed, and sorting of cells via an electric field (as for the FACS machine) or microfluidic pneumatic valves. We decided that this was very ambitious at the time. Precise and reproducible Raman measurements of individual cells is key to RACS, and the use of optical tweezers to immobilize cells during measurement, although slowing the process during manipulation, provided the necessary stability.
Our protocol is the third outcome1,2 of our collaborative project with the team at the University of Vienna, allowing us to present the platform and provide first analysis of functions within the mammalian intestinal microbiome. We were inspired to write the protocol by comments from a reviewer of our Nature Microbiology paper1, who outlined the need to make our platform broadly accessible to microbiologists lacking engineering expertise.
Novel systems can be developed in two ways: from scratch or by evolution based on integration of existing techniques. Integration of multiple components achieves far more than a simple sum of each, but requires substantial engineering to make them work together as meshed gearwheels (Fig. 1) – otherwise, they would slip. Over its history, humanity has been creating myriad groundbreaking technologies through this kind of integration. A recent representative example would be a smartphone. It is an integrated platform of existing technologies (telephone, camera, email app, web browser, music player, etc.), that has created revolutionary functionalities. In our case, integration was no easy task and it took countless iterations to integrate the five components of Raman microspectroscopy, optical tweezers, microfluidics, stable isotope probing, and operation software built in-house to achieve the novel functionality and smooth operability.
Fig. 1 | An integration of components creates novel functionalities. In our work, we have developed an automated platform for Raman-activated cell sorting (RACS) based on five major components, optical tweezers, Raman microspectroscopy, stable isotope probing, microfluidics, and operation software built in-house, yet their integration requires considerable peripheral engineering to give them functionality (image courtesy of Ms. Jihyun Kim).
Our platform opens a new avenue to establish RACS, but the technology has not yet reached its full potential. The throughput, which determines the number of cells that can be retrieved for large-scale applications like downstream omics or targeted cultivation of sorted cells, is one crucial aspect that can still be improved for wider applicability. There have been reports of higher throughput in a RACS platform based on coherent Raman microspectroscopy3. However, this throughput comes at a cost, for the use of coherent Raman lessens the great advantage of RACS over FACS – its ability to provide rich and diverse information concerning the molecules that constitute cells of interest – because it focuses on measurements of confined spectral regions. My perspective is that spontaneous (fundamental) Raman has great potential for RACS applications, whereas coherent Raman could be a game changer for imaging based on targeted spectral regions. There is room for further improvement of throughput using the spontaneous Raman system based on innovations in system configuration, the design of microfluidic devices, and methods to efficiently analyse multivariate Raman data. We are at a pivotal juncture in the application of Raman technology to microbiology. I am grateful to have been part of this adventure.
- Lee, K.S. et al. An automated Raman-based platform for the sorting of live cells by functional properties. Nature Microbiology 4, 1035–1048 (2019).
- Pereira, F.C. et al. Rational design of a microbial consortium of mucosal sugar utilizers reduces Clostridiodes difficile colonization. Nature Communications 11, 5104 (2020).
- Nitta, N. et al. Raman image-activated cell sorting. Nature Communications 11, 3452 (2020).