Recent developments in biomedical vibrational spectroscopy now permit the non-invasive imaging of cells and tissues within both the laboratory and clinical settings. The rapid nature and diagnostic potential of both Raman and FTIR (Fourier-transform IR) spectroscopy have resulted in their widespread application to a number of biological fields including fundamental cell biology, medical imaging, tissue engineering and pharmacology. In particular, Raman microspectroscopy shows tremendous promise for the analysis of biological processes within living cells, such as cell cycle dynamics, cell differentiation and cell death. Unlike conventional biological assays, laser-based Raman spectroscopy enables rapid and non-invasive biochemical analysis of cells in the absence of fixatives or labels. The low Raman signal of cell culture buffer/media permits the rapid monitoring of living cells growing under standard cell culture conditions. The Raman spectrum of a cell is a biochemical ‘fingerprint’, containing molecular-level information about all biopolymers contained within the cell. The high information content of Raman spectra can be used to characterize the distribution of multiple cellular components, and to study the dynamics of subcellular reactions, with excellent spatial resolution. This review highlights recent developments in Raman microspectroscopy, with a focus on non-invasive biochemical analysis of single living cells.
- biochemical ‘fingerprint’
- live cell analysis
- non-invasive cell biology
- Raman microspectroscopy
- single-cell analysis
An accurate understanding of cellular processes and responses to stimuli is of paramount importance in biomedical research areas such as cancer research, pharmacology and tissue engineering. Biologists have at their disposal a number of powerful techniques to characterize cells, populations of cells, organoids or even subcellular dynamics within single cells. A plethora of immunochemical, molecular biochemical and microscopy techniques have been developed to study cellular behaviour, based on probing of surface markers or intracellular content. New developments, such as total internal reflectance microscopy and fluorescence lifetime imaging microscopy, enable single-molecule imaging, allowing scientists to perform biochemical experiments inside living cells . Most techniques are invasive, requiring cell fixation, lysis, extraction or the introduction of molecular probes. Optical and spectroscopic methods, such as fluorescence or confocal microscopy, can be applied to the analysis of single living cells, but are disadvantaged because they require the use of exogenous fluorophores, and provide information limited to a small range of subcellular components. Laser-based Raman microspectroscopy presents a rapid, reagent-free and non-destructive alternative for the analysis of cell biology systems, and, in particular, single living cells.
Modern vibrational spectroscopy has developed into a powerful tool for biomedical research. The extension of microprobe and fibre optic technologies to spectroscopic imaging, as well as improvements in optical filter technology and detector sensitivity [2,3], have facilitated the widespread application of vibrational spectroscopy to the study of biological samples (proteins, cells and tissues) in the laboratory and clinical settings.
The greatest benefits of this technique are its high sensitivity and capability for non-invasive sensing . Thus unlike conventional biological assays, biochemical analysis of cells and tissues with Raman spectroscopy does not require the use of fixatives, markers or stains. It is a non-contact technique based on the inelastic scattering of monochromatic radiation (laser light). When a sample is irradiated, an exchange of energy takes place between the excitation light and the molecules of the sample, which results in a measurable shift in the wavelength of the incident laser light [3,5]. The resulting Raman spectrum is essentially a ‘biochemical fingerprint’, containing bands representing molecular normal modes of vibration of all molecules within the interrogated region of the sample . Since molecular moieties of all cellular biopolymers are simultaneously probed, the information content of Raman spectra is high. The coupling of a Raman spectrometer with an optical microscope, known as Raman microspectroscopy, facilitates the collection of spectra from volumes <1 μm3, enabling the analysis of micro scale features of biological samples. As such, Raman microspectroscopy can provide molecular-level information about the biochemical composition and structure of cells and tissues with excellent spatial resolution [7,8].
Raman scattering spectroscopy provides information similar to that of FTIR (Fourier-transform IR) absorption spectroscopy. However, the strong absorption of water in FTIR measurements hampers the analysis of hydrated biological samples. In contrast, Raman spectroscopy is better suited to the study of living cells, as spectra from aqueous media demonstrate only minimal interference from water. Raman spectroscopy can also offer superior spatial resolution than FTIR (typically ∼1 μm for Raman and ∼10 μm for FTIR [7,8]), due to the shorter wavelength lasers used for excitation [in the UV-, visible- and NIR (near IR)-range of the electromagnetic spectrum].
This review highlights recent developments in biological Raman microspectroscopy, with a focus on applications to non-invasive cell biology. The use of Raman imaging and microspectroscopy to monitor cellular events in living cells is discussed, along with potential benefits to biomedical applications.
Non-invasive cell biology with Raman microspectroscopy
Raman microspectroscopy has been used to examine cell populations in suspension [9–11], as well as single fixed , dried , cytospun  and living cells , grown or deposited on an appropriate Raman-transparent substrate (e.g. CaF2 and MgF2). Although interesting spectral information has been obtained from cell suspensions, single-cell spectra offer important information concerning specific biological functions, cellular interactions with its microenvironment (with other cells, drugs, biomaterials etc.), as well as variation within a population.
The Raman scattering effect is a weak process, and typically only 1 in 106–108 photons undergo an inelastic light scattering event . To facilitate the rapid analysis of single cells, dispersive Raman microspectrometers are configured with high-power NIR lasers and sensitive CCD (charge-coupled device) detectors [3,4] to promote maximal signal generation and efficient collection. This combination drastically reduces measurement times, and the use of NIR excitation avoids photodegradation commonly encountered with UV excitation, and reduces the fluorescence interference generated with visible-wavelength excitation. These features make dispersive NIR-Raman microspectroscopy a powerful tool for analysis of living cells growing in standard cell culture conditions (cell culture media/buffer at 37°C) [15–17].
Raman imaging and mapping of single cells
The distribution of biochemical species within a cell can be studied by directly obtaining a spatially resolved chemical image using a Raman microspectrometer. In DRI (direct Raman imaging), or wide-field imaging, the scattering of a Raman active vibration from a microscope FOV (field of view) is filtered and focused to produce an image directly on the detector [3,18]. The rapidity and high spatial resolution of DRI  make it a useful technique for imaging living cells in culture, as demonstrated by the imaging of β-carotene levels in live corpus luteum cells , and the subcellular distribution of an anticancer drug in living breast cancer cells . DRI has also been used to probe cell–biomaterial interactions, namely polymer microsphere degradation within fixed macrophages . A limitation of DRI is that only a narrow band of scattering is detected; therefore chemical resolution is restricted, and only a single component can be imaged at any one time. Furthermore, imaging requires special system configuration (tuneable filters for wavelength selection), as well as a priori knowledge of the Raman band for the target component.
Alternatively, a serial imaging technique can be used, whereby images are constructed from several point spectra collected at different spatial positions within a cell. In Raman microspectroscopic mapping, also known as hyperspectral imaging, a grid is defined on a microscope FOV, and a spectrum is collected at each pixel by moving the sample in a raster pattern with small step increments (Figure 1A). The laser can be focused to a diffraction-limited spot size (∼1 μm), but at such a high spatial resolution, spectra of subcellular organelles can be obtained with features dominated by only a single biochemical component (e.g. cholesterol [22,23] and phosphatidylcholine ). Thus point-to-point measurements may exhibit large spectral variation, and so automated collection of hundreds or thousands of spectra is required to completely characterize a cellular system. Since a complete Raman spectrum is associated with each pixel, such maps can be used to generate detailed chemical images revealing the distribution of a number of different components [7,23–25]. While individual measurements boast diffraction-limited spatial resolution, the overall spatial resolution is governed by the focused laser spot size and the accuracy of the motorized sample stage.
High spatial resolution Raman mapping can reveal in exquisite detail the structure and arrangement of subcellular organelles, and can provide insight into cellular biochemical dynamics. Recent studies reporting the subcellular detection of pyknotic nuclei in oxidatively stressed fibroblasts , and the distribution of condensed nuclear chromatin in human cells at different stages of mitosis , demonstrate the elegance of this approach. With this method, however, data collection time is in the order of several hours per cell (∼1–15 h or more) [7,23], and generating images from such large data sets (potentially hundreds of megabytes) is time-consuming and computationally intensive, requiring powerful computer processors and complex multivariate methods of spectral interpretation . High spatial resolution Raman mapping is therefore only suitable for the analysis of fixed and/or dried cells [7,23]. Fixing biological systems with alcohols or aldehydes is invasive, and can lead to distorted spectral features and artefacts . Cell fixation also eliminates two fundamental and attractive features of Raman microspectroscopy: non-invasive analysis and minimal sample preparation.
Spectral analysis of live cells
An alternative approach for live cell spectral analysis involves compromising between spatial resolution and spectral acquisition time. By employing a laser with a large spot size (not diffraction-limited), the scattering signal from a large subcellular region is represented in each Raman spectrum, and so only a small number (typically one to five) of measurements are required to map a cell. Compensation for the sacrifice in spatial resolution is the collection of high-quality spectral data from single living cells in only a few minutes. Rapid data acquisition minimizes possible interference from cellular changes, while simultaneously maximizing the number of cells sampled under the same in vitro conditions. This approach permits the analysis of a population of living cells in only a fraction of the time required for the collection of one high spatial resolution map of a single, fixed cell. In addition, analysis of the same cell population in a multiple time-point experiment is also possible.
Our group has developed a dedicated Raman microspectroscopic system optimized for the real-time analysis of single living cells . A high-power NIR-785 nm laser with an elliptical beam profile (∼10 μm×20 μm spot size with a ×63 water immersion objective) offers spatial resolution similar to that of diffraction-limited FTIR microspectroscopy , and allows rapid spectral acquisition from cells growing under standard cell culture conditions. Typical data collection time per cell is 5–10 min, depending on the integration time and the number of spectra required per cell, which in turn depends on cell type, size and morphology (Figure 1B). Spectra from different cellular locations can then be ensemble-averaged to produce a representative spectral biochemical ‘fingerprint’ of the cell. We have used this approach to examine a number of fundamental biological processes, including cell cycle dynamics , cell death , mRNA translation during differentiation of embryonic stem cells , and foetal osteoblast differentiation . Recently, this approach was used to monitor drug-induced apoptosis in single living cells (Figure 2) . Raman spectral analysis of lung cancer cells exposed to the anticancer agent etoposide reveals changes consistent with the drug's mechanism of action, namely the induction of double-stranded DNA breaks in dividing cells. Least-squares fitting (Figure 2B) of a set of basis components to cellular spectra was used to generate relative concentration profiles of major cellular components . The data indicate a significant reduction in DNA content and an increase in cellular lipids upon increasing exposure to the drug (Figure 2A). This example highlights the application of Raman microspectroscopy for in vitro toxicology evaluation of pharmaceuticals, as well as its ability to probe cellular processes within living cells. It also complements a recent high spatial resolution Raman mapping assessment of drug-induced apoptosis in fixed lung fibroblast cells . These approaches can provide a wealth of information about detailed subcellular reactions as well as overall live cell response.
Live cell phenotyping with Raman microspectroscopy
Live cell analysis with Raman microspectroscopy produces cellular ‘fingerprints’ that can be used to discriminate between different cell types. This approach has been used to differentiate between normal and transformed cells , cancer cells  and to detect different states within the same cell type (live compared with dead cells , cell cycle stage  or differentiation state ). The combination of Raman microspectroscopy with optical tweezers, LTRS (laser tweezers Raman spectroscopy), enables the acquisition of Raman spectra of cells in suspension . LTRS has recently been used for phenotypic discrimination between normal and neoplastic haemopoietic cells . The results show promise for the development of a non-invasive LTRS-based flow cytometry system for cell sorting, whereby sorting criteria are based on inherent cellular biochemical signatures rather than surface phenotype or fluorescent labels.
Non-invasive phenotyping with Raman microspectroscopy could have far-reaching applications, including identification of cancerous cell phenotypes to aid in disease detection and cancer research, as a biosensor to monitor cell response to drugs for pharmaceutical testing, and as a basic cytology tool to verify cell phenotype. Characterization of primary cells and maintenance of phenotype in culture are extremely important issues in cell biology, tissue engineering and stem cell-based therapies. It is common to use cancerous and transformed cell lines as models for primary cells, because they grow quickly and display more stable phenotype in culture. However, these cell lines display phenotypic differences to primary cells, and Raman microspectroscopy has been used to discriminate between primary human osteoblasts and cell line phenotypes used in bone tissue engineering . Some recent findings presented in Figure 3 reveal significant biochemical differences between primary human ATII (alveolar type II pneumocytes) and the model A549 adenocarcinoma cell line. Peaks in the difference spectrum (Figure 3C) indicate that intracellular lipids are primarily responsible for the differences, while PCA (principal components analysis) of the data (Figure 3D) revealed 100% spectral discrimination between clusters of ATII and A549 cells.
Enhanced and non-linear Raman techniques
A number of signal enhancement techniques with biological applications have been developed in addition to spontaneous Raman microspectroscopy. RRS (resonance Raman spectroscopy) has been used to image haemozoin within functional erythrocytes , and has been implemented with confocal imaging of live cells to identify subcellular components responsible for the uptake and metabolism of pharmaceuticals . The close proximity of an analyte to a metal substrate is required for SERS (surface-enhanced Raman spectroscopy), a method particularly useful for studying intracellular pharmacokinetics , and as a nanosensor to probe endosomal compartments within living cells . Multiple lasers can be used to excite coherent anti-Stokes Raman scattering [CARS (coherent anti-Stokes Raman spectroscopy)], a non-linear optical process. CARS microscopy is a rapid vibrational imaging technique with three-dimensional sectioning capability and has been used to image mitosis and apoptosis in live, unstained cells  and for real-time visualization of intracellular hydrodynamics . Enhanced Raman approaches do have limitations: they are technically more complex than spontaneous (non-resonant) Raman microspectroscopy; the UV- and visible-wavelength lasers used in RRS can result in high fluorescence interference or even sample photodegradation ; culturing cells on a metal substrate or in a colloidal solution for SERS can alter cellular behaviour ; and CARS suffers from limited chemical resolution [23,43]. Signal-enhanced Raman techniques and their biological applications are highlighted in a number of excellent review articles (RRS , SERS  and CARS [43,46]).
Raman microspectroscopy has proved to be a powerful analytical technique for biomedical applications. Its rapid, non-invasive and information-rich features make it an attractive modality to monitor biochemical dynamics and processes within single living cells. Optimization of Raman microprobe technology for cell biology platforms such as microarrays and multiple-well plates will encourage scientists to take advantage of automated analysis and high-throughput system capability. Correlation and validation of Raman spectral results with gold-standard biological techniques is essential if Raman spectroscopy is to reach its full potential. Future improvements in spectroscopic instrumentation and computing power will lead to shorter scan times, so that hyperspectral chemical imaging of single living cells may become possible. We expect that these and other developments will further enhance the suitability of Raman microspectroscopy for non invasive single-cell biology.
We thank the Rothermere Foundation, National Sciences and Engineering Research Council (Canada) and the Canadian Centennial Scholarship Fund for fellowship support to R.J.S.
Bionanotechnology: From Self-Assembly to Cell Biology: Biochemical Society Focused Meeting held at Homerton College, Cambridge, U.K., 3–5 January 2007. Organized and Edited by T. Cass (Imperial College London, U.K.) and D. Woolfson (Bristol, U.K.).
Abbreviations: ATII, alveolar type II pneumocytes; CARS, coherent anti-Stokes Raman spectroscopy; DRI, direct Raman imaging; FOV, field of view; FTIR, Fourier-transform IR; LTRS, laser tweezers Raman spectroscopy; NIR, near IR; PCA, principal components analysis; RRS, resonance Raman spectroscopy; SERS, surface-enhanced Raman spectroscopy
- © 2007 Biochemical Society