TECHNICAL NOTE pubs.acs.org/ac
Multidimensional Analysis of Single Algal Cells by Integrating Microspectroscopy with Mass Spectrometry Pawel L. Urban,† Thomas Schmid,† Andrea Amantonico,† and Renato Zenobi* Department of Chemistry and Applied Biosciences, ETH Zurich, CH-8093 Zurich, Switzerland
bS Supporting Information ABSTRACT: We demonstrate a facile label-free approach for performing multidimensional chemical analysis on individual single-cell organisms by combining optical, fluorescence, and Raman microspectroscopy with matrix-free laser desorption/ ionization mass spectrometry (MS). Single unicellular algae are seeded on a bare stainless steel plate and analyzed microspectroscopically. This provides information on the content and distribution of photoactive species, such as β-carotene, as well as chlorophyll and other components of the photosynthetic apparatus. Exactly the same cells are then analyzed by mass spectrometry in the negative ion mode. Phospholipid species are readily ionized by laser desorption/ionization of intact cells, without the need for an auxiliary matrix. This not only facilitates sample preparation but also preserves high spatial resolution and high sensitivity. Using this method, we were able to study the content and arrangement of proplastids and photosystem components, as well as the amounts of various phospholipid species in individual algal cells. The methodology can be used in the fundamental biological studies on these unicellular organisms, which require information on the internal structure as well as the chemical composition of individual cells.
A
vast number of bioanalytical protocols use samples containing a large number of cells. However, populations of cells, even those that have similar genetic constitution, can display substantial phenotypic variability.1 Single-cell analysis is essential to gain an unbiased insight on the biochemical processes that occur in heterogeneous populations of cells. Several methods for determination of different compounds in single cells have been presented to date.2,3 Moreover, it has been emphasized that a thorough understanding of single-cell systems requires multiparameter measurements, in which different components of the cellular phenotype are assayed.4 Integration of various orthogonal techniques is quite common in analytical chemistry. A synergy of different approaches yields information that could not be obtained by implementing individual techniques separately. An example of such study is the one in which image cytometry and capillary electrophoresis with fluorescence detection were used to correlate cell cycle with metabolism of single cells.5 However, conducting such multiparameter measurements on single cells is still challenging. Here, we present a simple method for obtaining multidimensional chemical data of single cells, which combines microscpectroscopic analysis and imaging with mass spectrometry. In an early work on single-cell mass spectrometry, a laser microprobe mass analyzer (LAMMA)6 was applied to detect inorganic cations in single mycobacteria.7 Since then, a number of studies on single-cell analysis using matrix-assisted laser desorption/ ionization (MALDI) mass spectrometry (MS) have been published (for reviews, see, for example, refs 8 and 9). Recently, it was also shown that, by utilizing a tightly focused laser beam in matrix-free laser desorption/ionization (LDI), it is possible to r 2011 American Chemical Society
visualize the distribution of naphthodianthrones and flavonoids in plant cells,10 and the potential of atmospheric pressure (AP) ionization MS approaches in this context has been successfully demonstrated in studies using AP-MALDI imaging for tissue analysis11 and a modified laser ablation electrospray ionization (LAESI) approach for single-cell analysis.12 In the present work, we propose a combination of matrix-free LDI-MS and microspectroscopic imaging of the intracellular matrix to obtain multidimensional chemical information for individual single-cell organisms. Such an integration of the two techniques brings important advantages: for a given cell, it is possible to obtain semiquantitative information on numerous phospholipid species directly desorbed from cellular membranes and, at the same time, acquire microscopic images revealing the presence and arrangement of different organelles (e.g., chloroplasts). Moreover, the images obtained by microspectroscopy bear chemical information on the intracellular distribution of pigments and other components of the photosynthetic apparatus. We believe that this combination of different analytical techniques, which yield complementary information, can make an important contribution to multiparameter analyses of single cells that have been identified as an important need in recent reviews (see, for example, ref 13). Our approach is label-free because we take advantage of the autofluorescence and Raman scattering of the analytes. In addition, ion formation in LDI-MS does not require sample preparation, for example, by adding a chemical Received: October 12, 2010 Accepted: December 23, 2010 Published: February 07, 2011 1843
dx.doi.org/10.1021/ac102702m | Anal. Chem. 2011, 83, 1843–1849
Analytical Chemistry matrix. A first attempt to analyze a single algal cell with MALDIMS as well as Raman and fluorescence microscopy has recently been shown by our group in ref 14.
’ EXPERIMENTAL SECTION Cell Culture and Sample Preparation. A culture of Euglena gracilis cells was purchased from Carolina Biological Supply Company (Burlington, NC, USA). They were maintained at room temperature and illuminated for 16 h during a day/night cycle with an electric bulb (13 W, 900 lm; Osram Dulux, Munich, Germany). Before the measurement, a small quantity of the culture (typically, 1 mL) was centrifuged at 1000 rcf for 5 min and the pellet was resuspended in deionized water. Microscopic observation revealed that the cells were still active/motile in pure water. Cells were seeded on sample posts of a standard stainless steel MALDI plate (AB Sciex, Concord, ON, Canada). Excess medium was removed with a paper tissue. Attention was paid to achieve spatial separation of cells and limit cell aggregation. Raman and Fluorescence Microspectroscopy. The first step in the optical/microspectroscopic part of this study was the observation of the MALDI plates using a white-light stereomicroscope (Nikon) and the collection of images from single spots on the plate revealing the spatial distribution of the cells. On the basis of the steromicroscope image, we are able to assign mass-spectrometric and microspectroscopic measurements to exactly the same cell. In the next step, the plate was placed below the objective of an NTegra SPECTRA system (NT-MDT; Zelenograd/Moscow, Russia), as described in ref 15, with a 532 nm diode-pumped solid-state laser (DPSS) employed for excitation. Since the laser-scanning instead of the sample-scanning mode was used for microspectroscopic imaging, the plate could be placed there without any special mounting (for details about the operation of the instrument, see ref 16). The whitelight microscopy module of the system was used together with micrometer screws at the sample support to locate single cells inside one spot of the MALDI plate. Subsequently, the system was switched to laser-scanning mode, and a confocal image of a selected cell was recorded. This allowed an exact overlap of the optical white-light image of the cell with microspectroscopic images recorded later on the exact same place. Comparison of cell shape and surface structures of the MALDI plate on both kinds of images enabled an exact overlap. Then, the laser was raster-scanned over the selected area of the sample, and the whole spectrum was collected in every pixel. Microspectroscopic mappings were performed on sample areas of 35 35 μm2 in size with a resolution of 64 64 pixels or 550 nm per pixel, respectively. This corresponds approximately to the optical resolution of the system. The strong fluorescence emission and Raman scattering, which is resonantly enhanced at this particular wavelength, in combination with the high collection and detection efficiency of the system allowed the use of a laser power as low as ∼5 μW at the sample and an acquisition time of only 0.1 s per pixel. Mass Spectrometry. We used a commercial MALDI-MS instrument (model 4800 Plus; AB Sciex) equipped with a solid-state laser with a wavelength of 355 nm. The effective diameter of the laser beam focus was ∼30 μm. For surface mapping, we chose a 2 2 mm raster, covering the whole sample post. The laser power used for LDI scans was ∼50% higher than the optimum laser power used for MALDI with matrices such as 2,5-dihydroxybenzoic acid (6250 against 4100 in relative units of
TECHNICAL NOTE
the data acquisition software). For the analysis of Euglena gracilis cells, a raster spacing of ∼36 μm was chosen. In fact, the lateral resolution of our LDI-MS experiment was not far behind stateof-the-art MS imaging and was sufficient to record one mass spectrum for each individual cell. At each position, 25 laser shots were fired, and one spectrum was collected. Peak areas were exported using the analysis tool in the 4000 series software (AB Sciex) and further treated using MATLAB (ver. 7.6.0.324 (R2008a); MathWorks, Natick, MA, USA).
’ RESULTS Microspectroscopy. Figure 1 shows microscopic and microspectroscopic analysis of single cells of Euglena gracilis, the same ones as those presented in Figure 2A (cells 1 and 2). Figure 1A-H allows a direct comparison of both cells based on optical microscopy, Raman, and fluorescence images. The whitelight microscopy images in panels A and E reveal the shape and size, as well as internal structures of the cells. The “greenish” color is due to chlorophyll pigments. The intensity of the Raman band at 1515 cm-1 (see spectra in Figure 1M) is a measure for the β-carotene concentration, and Figure 1B,F shows the distribution of this pigment inside the cells. For obtaining these maps, fluorescence background at the left and right of this band in the spectrum was subtracted and only the Raman intensity was plotted. The reference spectrum in Figure 1M (trace 1) of pure β-carotene powder (Fluka) confirms the assignment of the bands at 1150 and 1515 cm-1 to this pigment. Since literature values of Raman shifts can show slight variations due to the use of different spectrometers, spectral resolutions, and laser wavelengths as well as deviations in spectrometer calibration, we measured a reference spectrum of this carotenoid with the same instrument settings for an unambiguous band assignment. Fluorescence emission at 618 nm (cf. Figure 1M) was assigned to a chlorophyll pigment that is present in proplastids,17 and the intensity distribution at this specific wavelength visualizes these small globular organelles and their distribution inside the cells (Figure 1C,G). We would like to emphasize that spectra 2 and 3 in Figure 1M were not artificially shifted along the intensity axis. The overall increased intensity of spectrum 3 compared to spectrum 2 represents the chlorophyll fluorescence band that is obviously broader than the spectral range shown here. We chose the indicated range around 618 nm as the basis for the fluorescence maps, because we got the best contrast for imaging of proplastids in this spectral region. There is a relatively sharp band in the spectrum superimposed on the broad fluorescence emission at 618 nm. This can be assigned to an overtone of the main Raman bands of β-carotene (compare the reference spectrum in Figure 1M, trace 1). Due to its low intensity, a contribution of this band to the fluorescence maps in Figure 1C,G can be neglected. Proplastids are formed in Euglena gracilis at the beginning of their cell cycle and are subsequently transformed into photosynthetically competent chloroplasts containing different kinds of chlorophylls that emit in different ranges of the electromagnetic spectrum. These chloropasts are shown for cell 2 in Figure 1I,K, which reveal the distribution of fluorescence intensities that can be attributed to the reaction center (RCII) of photosystem II (Figure 1I, emission at 675 nm; see spectra in Figure 1N) and internal antennae of photosystem II (PSII, Figure 1K, emission at 695 nm). Figure 1D,H overlays β-carotene distribution (red) with proplastids (green), and in Figure 1L, 1844
dx.doi.org/10.1021/ac102702m |Anal. Chem. 2011, 83, 1843–1849
Analytical Chemistry
TECHNICAL NOTE
Figure 1. Optical microscopy (A, E) and microspectroscopic images (B-D, F-L) of Euglena cells 1 (A-D) and 2 (E-L) that were also studied by LDIMS. The scale bar in (B) applies to all images. The microspectroscopic images are based on the intensities of specific bands in Raman (M) and fluorescence (N) spectra of the cells, where black corresponds to the lowest intensity. (M2, 3) and (N) show representative spectra taken from microspectroscopic mapping experiments of cell 2. (B and F) show the distribution of β-carotene in the cells based on Raman spectroscopy. In (C and G), fluorescence emission from proplastids is visible, and (D) and (H) overlay the β-carotene (red) and proplastid distribution (green). (I-L) show additional information about cell 2 obtained by fluorescence microspectroscopy revealing the distribution of pigments related to the reaction center RCII (I) and internal antennae of photosystem II (PSII; K). (L) overlays the distributions of pigments related to RCII (blue) and internal antennae of PSII (green).
pigments related to RCII (blue) and internal antennae of PSII (green) are shown. Assignment of fluorescence emissions at certain wavelengths to specific chlorophyll pigments is based on ref 17 and references cited therein. Photosynthetic pigments emitting in the spectral range around 700 nm are often termed “low energy” or “red” chlorophylls.18 Structure and stoichiometry of protein complexes that are involved in this fluorescence emission are targets of ongoing research.19 In general, Figure 1 demonstrates the high information content of Raman scattering and autofluorescence emission of Euglena gracilis cells. In contrast to the collection of micrographs that are based on the overall autofluorescence of cells (see, for example, ref 17), acquisition of the whole spectrum in every pixel allows one to take advantage of both, Raman scattering and autofluorescence emission, where in the latter case all possible
fluorescence channels are recorded simultaneously. This reveals the distribution of several different pigments inside the cell and enables the investigation of colocalization of different substances. Thus, full spectroscopic maps of single algal cells allow deeper insight into the highly complex processes related to photosynthesis. A comparison of Figure 1F,K, for example, speaks for a colocalization of β-carotene and the plastids containing internal antennae of PSII. It should be pointed out that the fluorescence and Raman measurements were performed using a confocal microscope, whose focal plane was aligned to the level of the steel substrate. Thus, all microspectroscopic images in Figure 1 reflect the distribution of the mentioned pigments inside an ∼1 μm thick layer of the cell close to its interface with the substrate. In this proof-of-concept study, algal cells were chosen as a target for several reasons (see Concluding Remarks). One of 1845
dx.doi.org/10.1021/ac102702m |Anal. Chem. 2011, 83, 1843–1849
Analytical Chemistry
Figure 2. Negative ion mode laser desorption/ionization mass spectra of single unicellular organisms (Euglena gracilis). (A) Optical view and selected ion maps of the surface of stainless steel plate with dispersed single cells. Pixel ensembles in the ion maps (m/z 793 and 467) generally correspond to the positions of the cells as visualized by stereomicroscope. White ellipses mark cells 1 and 2. (B) LDI mass spectra acquired for the two cells marked in (A), consistent with ions derived from phospholipids. (C) MS/MS spectrum of the ion at m/z = 793.5 obtained in a separate experiment with multiple cells. The insets show zoomed views of the mass spectra.
them was their spectroscopic activity, i.e., intense autofluorescence emission of chlorophylls and resonantly enhanced Raman scattering of β-carotene, leading to very short accumulation times of 0.1 s per pixel. In principle, Raman microscopy can also be applied to nonpigmented cells. For example, studies of nonpigmented bacteria have shown the high potential of this technique, which yields spectroscopic fingerprints of lipids, proteins, and other cell constituents.20 On the basis of literature values, we expect measurement times in the order of 1 min per pixel for nonpigmented cells due to the much lower Raman scattering cross sections of these compounds.
TECHNICAL NOTE
LDI-MS. Figure 2 presents data collected during LDI-MSaided surface mapping of Euglena gracilis cells, which were randomly spread over a stainless steel support. The positions of the cells visualized by a stereomicroscope (optical image) generally match the positions of the pixel ensembles in the MS maps (Figure 2A). These MS maps provide chemical information on the individual cells via two prominent peaks assigned to phospholipids. Variations of the relative phospholipid contents among the cells can easily be spotted. For example, one of the cells (cell 3) shows a high level of the species represented by the MS peak at the m/z 793, while the content of the species represented by the MS peak at the m/z 467 is considerably lower. The individual single-cell mass spectra exhibit a large number of peaks revealing richness of the phospholipid composition of the cell (Figure 2B). In general, the spectra show two characteristic groups of peaks, in the m/z ranges 400-600 and 750-850. When the two spectra corresponding to cells 1 and 2 (cf. Figure 2A, left panel) are compared, it is easy to note different relative intensities of the peaks between these two groups (Figure 2B). In a similar way, it was possible to obtain an LDI mass spectrum corresponding to an individual cell of Chlamydomonas reinhardtii (see Supporting Information). The chemical nature of the ions detected by the time-of-flight (TOF) mass spectrometer was verified on the basis of the following: (i) matching the exact masses of the phospholipids listed in the HMDB database (www.hmdb.ca), (ii) peak pattern in the mass spectra (Figure 2B) that is characteristic of lipid species, (iii) the fact that phospholipid species are readily ionized by LDI in negative ion mode (see Supporting Information), (iv) as the final confirmation, analysis by tandem TOF/TOF mass spectrometry with collision-induced dissociation, which gave characteristic fragments (e.g., PO3-, m/z = 79; C3H6O5P-, m/z = 153; C16H31O2-, m/z = 255; Figure 2C). Figure 3 demonstrates the effect of irradiation of a single Euglena gracilis cell with the laser beam. The microspectroscopic evaluation of the cell residue indicates substantial depletion of the organic components of the cell: before the LDI experiment, characteristic Raman bands of β-carotene are found in the spectra, whereas after irradiation with the MALDI laser beam, only nonspecific fluorescence can be detected, which is most probably due to traces of decomposition products of organic matter.
’ DISCUSSION Phospholipids are frequently analyzed by MALDI-MS in positive ion mode, in cationized form.21 Several methods for detection of phospholipids by LDI-MS, sometimes aided by “smart” materials (e.g., nanostructured silicon,22 colloidal graphite,23 nanostructured diamond coating24) have also been reported. However, lipid species, for example triacylglycerols in olive oil,25 can also be detected without any special sample support. Building on this, we found that matrix-free LDI-MS, run in negative ion mode, provides sufficient sensitivity to detect phospholipid species in single algal cells of Chlamydomonas reinhardtii (see Supporting Information) and Euglena gracilis (Figure 2). The matrix-free approach, using unmodified stainless steel support, facilitates and speeds up sample preparation. In other work, Ishida et al.26 analyzed lipids in small animals with a diameter of around one up to a few millimeters (Daphnia galeata) using MALDI-MS with 2,5-dihydroxybenzoic acid as a matrix; this strategy can address issues related to the different extraction 1846
dx.doi.org/10.1021/ac102702m |Anal. Chem. 2011, 83, 1843–1849
Analytical Chemistry
TECHNICAL NOTE
Figure 3. Optical microscopy images (A, D), Raman intensity maps of the carotenoid band at 1515 cm-1 (B, E), and Raman/fluorescence spectra (C, F) of cell 1 taken before (A-C) and after the LDI-MS experiment (D-F). Images (B and E) are based on the same intensity scale that is normalized to the most intense pixel in the fluorescence map (E). Most parts of the cell in (B) appear white, because the signal is saturated on this color scale (see Figure 1B that is the same Raman map shown with a different color scale). The images prove complete disintegration of the cell and almost complete removal of the organic material from the plate by LDI. The spectrum taken after LDI shows nonspecific fluorescence emission that is significantly different from all spectra collected from intact cells and is most likely due to decomposition products of organic matter.
efficiencies of individual lipid classes.21 In LDI-MS, the analytes are not dispersed over a large area, as it might happen when applying chemical matrixes dissolved in some organic solvents: the analytes remain confined within the micrometer-scale semidry cell residue, typically smaller than, or comparable in size, with the laser beam focus. Spectral mapping of phospholipids in tissue samples can easily be achieved by MALDI imaging (e.g., refs 27 and 28); for example, using the 2,5-dihydroxybenzoic acid matrix applied in the form of fine powder, various phospholipid species could be recorded in the positive ion mode.27 Matrix-free imaging of phospholipid distributions in tissues can also be conducted by desorption electrospray ionization (DESI) MS:29 similarly to DESI, the matrix-free LDI method has the advantage of little or no sample preparation, ease of implementation, and high quality analytical result. However, the spatial resolution achieved with matrix-free LDI is considerably better than in DESI, which also helps to maintain a high signal-to-noise ratio (Figure 2B). An alternative method developed for chemical imaging of unicellular algae was presented by Hirschmugl et al.:30 synchrotron-based infrared spectroscopy was used to conduct chemical analysis and imaging of single Euglena gracilis cells. While Hirschmugl et al.30 obtained images for pools of various analytes with a resolution of 5 μm, here, we combine the high microspectroscopic resolution (
