Recent Advances in Single-Cell Metabolomics Based on Mass Spectrometry

Open AccessCCS ChemistryMINI REVIEWS22 Oct 2022Recent Advances in Single-Cell Metabolomics Based on Mass Spectrometry Qinlei Liu, Sandra Martínez-Jarquín and Renato Zenobi Qinlei Liu Department of Chemistry and Applied Biosciences, ETH Zurich, CH-8093 Zurich , Sandra Martínez-Jarquín Department of Chemistry and Applied Biosciences, ETH Zurich, CH-8093 Zurich and Renato Zenobi *Corresponding author: E-mail Address: [email protected] Department of Chemistry and Applied Biosciences, ETH Zurich, CH-8093 Zurich https://doi.org/10.31635/ccschem.022.202202333 SectionsAboutAbstractPDF ToolsAdd to favoritesDownload CitationsTrack Citations ShareFacebookTwitterLinked InEmail Cellular heterogeneity is essential for the physiological functions of organisms, and precise interpretation of the relevant biological mechanisms involved requires accurate measurement of biomolecules, including DNA, RNA, proteins, and metabolites at the single-cell level. Cellular metabolites react most rapidly to environmental and biochemical changes such that their detection is a critical step in determining the physiological state of cells. However, their limited amounts, structural diversity, and significant content variation render single-cell metabolomics analysis technically challenging. Mass spectrometry, with its high sensitivity, good selectivity, and wide dynamic range, has been widely used in the single-cell analysis in recent years. In terms of ionization techniques, there are four main types of single-cell mass spectrometry methods: nano-electrospray ionization mass spectrometry, matrix-assisted laser desorption/ionization mass spectrometry, secondary ion mass spectrometry, and dielectric barrier discharge ionization mass spectrometry. In this mini review, the progress of single-cell mass spectrometry methodologies and data processing strategies are summarized, and the future trends in single-cell mass spectrometry are outlined. Download figure Download PowerPoint Introduction Cellular heterogeneity exists between individual cells due to differences in intracellular gene expression levels, cellular physiological environment, cell age, and other factors that influence the phenotype of individual cells in culture.1 Identifying these differences can reveal critical information. Traditional population measurements obscure information on intercellular heterogeneity; thus, it does not yield the actual chemical composition and content of individual cells.2 In addition, although rare cell types such as circulating tumor cells (CTCs) or cancer stem cells play significant roles in pathogenic mechanisms and early diagnosis of tumors, due to their low abundance, bulk analysis hardly provides meaningful information for the study of rare cell types.3 Single-cell analysis, on the other hand, can deliver a wealth of information for the studies of cell physiology and pathology. Various single-cell "omics" approaches have been developed, including single-cell genomics, transcriptomics, proteomics, and metabolomics.4–6 Compared to other single-cell "omics" methodologies, metabolomics provides the most rapid and dynamic information on cellular functions, owing to the dynamic response of the metabolome to the environment on a short time scale (seconds or less), its structural diversity, and the extended dynamic range of metabolites.7–9 Single-cell metabolomics focuses on the determination of small-molecule metabolites with molecular weights less than ∼2 kD in specific cells, organs, and organisms, mainly including endogenous and exogenous small molecules (e.g., lactate, sugars, adenosine monophosphate, adenosine diphosphate (ADP), adenosine triphosphate (ATP), drugs, and their metabolites and lipids).1 However, single-cell metabolomics is challenging because of the low levels of single-cell metabolites; for example, the volume of a single Escherichia coli cell is around 1 fL, and the intracellular metabolite content is within fg level, even for metabolites of central metabolism that are present in high concentrations.10 Furthermore, intracellular metabolites cannot be amplified and subsequently detected, as is possible for intracellular DNA. Therefore, single-cell metabolomics requires highly sensitive and selective analytical techniques. In addition, in order to increase the coverage of single-cell metabolomics analyses, quantitative or semiquantitative analysis of as many cellular metabolites as possible is desirable. The emergence of diverse single-cell analysis techniques has greatly promoted the development of single-cell research. The main techniques currently used for single-cell analysis include electrochemistry,11–13 fluorescence microscopy,14–16 vibrational spectroscopy,17–19 capillary electrophoresis (CE),20,21 mass spectrometry,22–26 and so on. Mass spectrometry is a well-suited assay for single-cell analysis, with high versatility, good sensitivity, fast analysis speed, wide dynamic range, and the ability to analyze thousands of compounds simultaneously. Since mass spectrometry detects the molecular weight of each compound, mass spectrometry-based single-cell analysis methods are universal and do not require fluorescent labeling or derivatization; hence, extensively used in single-cell metabolomics.27–29 This mini review summarizes single-cell mass spectrometric methods from the perspective of the ionization method. We will discuss the three widely used ionization sources used for this purpose, nano-electrospray ionization (nanoESI), matrix-assisted laser desorption/ionization (MALDI), and secondary ion mass spectrometry (SIMS). In addition, this mini review also addresses for the first time the possibility of using dielectric barrier discharge ionization (DBDI) for single-cell analysis. Moreover, data analysis strategies used for single-cell mass spectrometry studies are discussed, and an outlook on the future trends of single-cell analysis is given. Ionization Techniques Used for Single-Cell Metabolomics Single-cell mass spectrometry methods based on different ionization techniques have been developed for single-cell metabolomics, including nanoESI-based methods,24,30,31 MALDI-based methods,32–34 SIMS-based methods,26,35,36 and DBDI-based methods.37,38 These ionization methods have their own characteristics and can complement each other. Below, the corresponding mass spectrometric single-cell analysis methods are described for each of these four ionization techniques. Electrospray ionization-based methods Electrospray ionization (ESI) is a "soft" ionization source with high ionization efficiency and has a wide range of applications in the life sciences.39 NanoESI offers higher ionization efficiency and sample utilization compared with traditional ESI technology. NanoESI-based single-cell analysis methods generally use a probe or laser for sampling.24,40 Some very interesting methods are based on the direct aspiration of liquid from certain parts of living cells using a microcapillary that can be moved with a manipulator in three-dimension (3D) with real-time observation under a microscope. After sampling, direct ionization can be performed by nanoESI. Masujima and coworkers24,41–43 have done pioneering work in this field, developing a "live single-cell MS" (live MS) method and applying it to analyze differences in the distribution of subcellular metabolites in the rat leukemia cell line, RBL-2H3 (Figure 1a).24 Subsequently, they successfully applied live MS for the analysis of metabolites in plant cells24,42 and CTCs.41 This early research using live single-cell MS has inspired many subsequent studies. For example, Yang and coworkers44 developed a single-probe-based nanoESI source to enable real-time, in-situ single-cell metabolomics analysis, whereby they successfully detected several cellular metabolites and low molecular weight (MW) anticancer molecules, including paclitaxel, doxorubicin, and orsaponin in single cervical cancer (HeLa) cells. To achieve real-time analysis of suspended live single cells, they further developed a T-shaped probe and realized the analysis of single colon cancer cells in suspension.45 Using this approach, it was possible to track how the metabolites in HCT-116 cells changed after treatment with anticancer drugs. To analyze metabolites at the subcellular level, Zhang and coworkers46 used a tungsten probe as a solid-phase microextraction (SPME) probe to directly sample and enrich metabolites in living cells, followed by desorbing/ionizing the enriched metabolites from the tip of the tungsten probe for MS detection. They successfully detected different types of metabolites, including six fructans, four lipids, and eight flavone derivatives in single Allium cepa cells. To demonstrate the subcellular detection capability of this method, they used a nucleus dye (methylene blue) to stain onion epidermal cells. The stained nuclei and cytoplasm were then sampled and measured separately by ESI-MS, and it was found that the signal intensity of methylene blue in the nucleus was approximately 20× higher than that in the cytoplasm. In order to perform in situ single-cell analysis and to overcome the severe ion suppression and matrix effects from the cytoplasm, Huang and coworkers47–50 developed induced nanoESI (InESI) for single-cell metabolomics. By combining InESI-MS with a patch-clamp electrophysiological approach, they successfully analyzed the physiological activity and metabolite composition of a single neuron.50 In general, probe-based sampling methods are direct and can be performed in situ, but the complicated sampling operation limits the throughput and reproducibility of this method; thus, only some metabolites can be detected in individual cells. Figure 1 | Examples of ESI-based single-cell mass spectrometry. (a) Schematic of live plant single-cell MS. Reprinted with permission from ref 24. Copyright 2015 Springer Nature. (b) Microprobe single-cell CE-ESI-MS enabling in-situ metabolic characterization of live X. laevis embryos. Reprinted with permission from ref 55. Copyright 2017 American Chemical Society. (c) Schematic of LAESI (top view) with plume collimation (left) and side view of conventional LAESI (right). Reprinted with permission from ref 61. Copyright 2013 American Chemical Society. (d) Schematic of a CyESI-MS detection system. Reprinted with permission from ref 65. Copyright 2019 American Chemical Society. Download figure Download PowerPoint Due to the complexity of the intracellular metabolome, nanoESI has also been coupled with separation techniques to reduce interferences between different compounds and improve detection sensitivity and coverage. CE is particularly well suited because only very small injection volumes of liquid samples are required.51–54 Nemes and coworkers55,56 applied CE-ESI-MS to the metabolomic analysis in single cells in developing Xenopus laevis embryos (Figure 1b),55 obtaining a limit of detection of ∼5–10 nM (50–100 amol) for dual cationic–anionic analysis.56 More recently, they achieved both single-cell proteomic and metabolomic analyses in X. laevis embryos by CE-ESI-HRMS. After in situ microprobe sampling, they tested the embryos' viability and found that they could still develop freely into visually normal-behaving tadpoles.57 Although still on the slow side in terms of throughput, CE-ESI-MS reveals a wealth of metabolite information in single cells, paving the way for a better understanding of cell metabolism. Ion-mobility separation (IMS) is another powerful separation platform used in single-cell metabolomics, which enables the separation of isobaric ions to enhance molecular coverage while decreasing chemical background within milliseconds.58–60 For example, Vertes and coworkers58 applied ESI-IMS-MS for a metabolic analysis of three types of A. thaliana epidermal cells (pavement cell, basal cell, and trichome) at the single-cell level. They discovered significant metabolic differences between the trichomes and the other two types of cells, which aided comprehension of plant biology and biochemistry at the single-cell level.58 In addition to probe-based sampling methods, laser-based sampling approaches have been developed for nanoESI-based single-cell analysis. For example, laser ablation electrospray ionization (LAESI) is a typical laser-based method that utilizes an IR laser with a specific wavelength for desorption and ESI for post-ionization. Vertes and coworkers40,61,62 proposed an in situ analysis of individual cells by LAESI-MS at atmospheric pressure. For example, they analyzed the metabolites in A. cepa, Narcissus pseudonarcissus bulbs, and Lytechinus pictus at the single-cell level; they also analyzed individual epidermal cells from different scale leaves in A. cepa bulbs simultaneously to study metabolic differences.40 Generally, LAESI uses an optical fiber to deliver the laser beam and is able to analyze large, single cells with an average volume of ∼1 nL. However, only a small fraction of the samples are ionized due to the expansion of laser-ablated gaseous samples in three dimensions, which becomes very challenging for the analysis of small cells. Thus, Vertes and coworkers61 utilized a capillary to limit the expansion of the ablation plume, improving its overlap with the electrospray plume, which achieved higher ionization efficiency and lower detection limits compared with conventional LAESI (Figure 1c). Using this method, they measured hexose, disaccharide, and a trisaccharide in single buccal epithelial cells (with diameters of 30–50 μm, volumes of 14–65 pL/cell), and glycine, trimethylamine dimer, and lipids in individual sea urchin eggs (with diameters of 90–100 μm, volumes of 400–500 pL/cell). These cells are larger than typical animal cells (with diameters of typically 10–30 μm), that is, LAESI still needs to be improved in terms of sensitivity for single-cell detection. An isolated measurement of a single cell or low-throughput measurements of several cells may have less biological significance, as hundreds or thousands of measurements are required frequently to generate statistically significant data in biology. Microfluidics-based techniques are effective for studying complex biological systems because of their low sample consumption, outstanding performance in handling trace liquids, and high throughput.63 The channel size of a typical microfluidic device is around 10–100 μm, which is close to the size of cells, and thus, operation ranges and processes at the single-cell level can be performed efficiently in microfluidic chips.64 To enable high-throughput analysis and to obtain biologically meaningful data; microfluidic techniques have been combined with nanoESI for single-cell analysis.60,65–68 For example, Zhang and coworkers65,66 proposed a capillary microfluidics platform-based flow cytometry ESI-MS (CyESI-MS) technique to achieve high-throughput single-cell analysis (38 cells/min), cell type discrimination, and biomarker screening (Figure 1d). Liu and coworkers67 combined nanoESI with microfluidic chips to identify 100 metabolites and six protein biomarkers in single cancer cells with a high throughput of 40 cells/min and successfully distinguished different cell types based on the information obtained for the simultaneous detection of proteins and metabolites. While floating cells are of interest for fundamental studies, most cells in multicellular organisms are organized in tissues. Therefore, analyzing tissue-embedded single cells in their natural environment with high throughput is highly desirable. Vertes and coworkers69 developed a bimodal microscopy imaging system along with fiber-based LAESI (f-LAESI)-MS with enhanced throughput (120 cells/h) for the ambient analysis of tissue-embedded single cells (n > 1000) to gain insight into cellular heterogeneity. These methods open the door for higher throughput single-cell analyses to obtain statistically significant data. MALDI-based methods MALDI is a soft ionization method used for investigating large molecules. MALDI-MS measurements require mixing the sample with a matrix and then irradiating the sample with an ultraviolet (UV) laser beam, usually in a high vacuum. The analytes are then desorbed and ionized, then enter the mass spectrometer for detection. MALDI has been widely used in the field of single-cell analysis due to its high spatial resolution and sensitivity.25,26,32,35,70–73 It is sensitive, fast, and provides simple mass spectra of the analytes even from complex mixtures. MALDI is a powerful tool for lipid analysis in particular. Sweedler and coworkers74 used MALDI Fourier-transform ion cyclotron resonance (FTICR)-MS to analyze over 30,000 individual rat cerebellar cells and t-distributed stochastic neighbor embedding (t-SNE) to distinguish different cell subtypes (Figure 2a). In this work, FTICR-MS provided high mass resolution and mass accuracy, allowing 500 brain lipids to be assigned without needing MS/MS analysis. This is probably one of the highest numbers of cells analyzed in a single-cell MS experiment. In addition, they combined MALDI-MS and immunocytochemistry (ICC) to analyze lipids in astrocytes and neurons, revealing the lipid heterogeneity between these two cell types.72 Although MALDI-MS enables high-throughput analysis of lipids for a variety of cells, it is usually not quantitative, and many low-molecular-weight metabolites are not detected due to interferences from the MALDI matrix. Sweedler et al.,70 therefore, coupled MALDI with CE-ESI-MS to effectively increase the throughput of single-cell analysis and improve the coverage of metabolites. Figure 2 | Examples of MALDI-based single-cell mass spectrometry. (a) Single-cell MALDI-MS for lipid analysis. Reprinted with permission from ref 74. Copyright 2019 American Chemical Society. (b) Graphical summary of the workflow used to prepare samples for single-cell MALDI-MS analysis. Reprinted with permission from ref 77. Copyright 2013 National Academy of Science. (c) Schematics of the MALDI-2 ion source. Reprinted with permission from ref 81. Copyright 2015 The American Association for the Advancement of Science. Download figure Download PowerPoint To simplify the sample preparation process and improve the throughput of single-cell analysis, our group developed a high-density microarray for mass spectrometry (MAMS) chips that can automatically generate picoliter (pL) aliquots that contain single cells from a cell suspension with the appropriate density (Figure 2b).75–77 For example, we studied the ATP metabolism and glycolytic metabolic pathway in single yeast cells with MAMS-MALDI-MS; we also compared the results of the single-cell analysis with those from population measurements.76,77 Although the trends were consistent, there was a significantly higher variability in the results of the single-cell analysis not found in the population measurements, as expected. To obtain multilevel information on single cells, we also combined MALDI-MS with Raman and fluorescence microspectroscopy to monitor the growth of single cells of green algae.71,78 These three methods applied to the same cells yielded complementary information and successfully identified metabolites such as ATP, ADP, astaxanthin, chlorophyll, and β-carotene. Mass spectrometry imaging (MSI) is a powerful and label-free technique for the untargeted investigation of the spatial distribution of metabolites in single cells, and MALDI is the most useful and widespread ionization source type for MSI.79 Recently, Alexandrov and coworkers80 developed SpaceM, an open-source platform that combines MALDI-MSI with light microscopy for in situ single-cell metabolomics. With the ability to detect over 100 metabolites per hour from over 1000 different cells, SpaceM provides an exciting tool for future single-cell metabolomics studies. Although MALDI-MSI is promising for the chemical imaging of cell membranes, the ionization efficiency of MALDI is low. To address this problem, Dreisewerd and colleagues81 developed MALDI-2 to enhance the signal for membrane components other than some highly concentrated lipids via a post-ionization strategy (Figure 2c). In contrast to prior photoionization investigations performed in high vacuum (p ≤ 10−6 mbar),82,83 MALDI-2 uses a pulsed UV laser beam that intercepts the expanding particle plume in an N2 cooling gas environment.84,85 By beam shaping and placing the focusing lens inside the MALDI ion source, the effective diameter of the primary laser beam reached 5 μm. The technology has improved the sensitivity of some molecules such as drug compounds86 and lipids,87 and it recently achieved a pixel size of 600 nm with brain tissue by employing transmission-mode MALDI-2 MSI.73 This technique could be a new valuable tool for cell biology and biomedical research. SIMS-based methods SIMS is a surface analysis technique with high sensitivity and spatial resolution, which usually uses a high-energy primary ion beam to bombard the sample surface and sputter molecules off the sample surface to generate secondary ions that are characteristic of the chemical composition of the sample (Figure 3a).88,89 Although SIMS is a relatively "hard" ionization source compared to MALDI and ESI, the greatest advantage of SIMS over other ionization sources is its extremely high spatial resolution, which can reach down to a few 10 s of nm. SIMS is capable of analyzing small molecules such as metabolites, lipids, and peptide fragments. Therefore, SIMS has been widely used for subcellular analysis and MSI.26,35,36 For example, the Ewing group90,91 has pioneered many SIMS imaging approaches for subcellular analysis. They used time-of-flight (TOF)-SIMS imaging to investigate changes in membrane structure driving the formation of lipid structural domains in mating Tetrahymena thermophila, and the high sensitivity and resolution of SIMS contributed to the understanding of domains in cell membranes. In addition, they applied SIMS to image the fly brain and demonstrated that the drug methylphenidate alters lipid composition. To find a correlation between the chemical and morphological structure of the brain, they used scanning electron microscopy imaging along with SIMS imaging.92 This discovery helped to clarify the relationship between biomolecular distribution and brain function. They also combined nanoSIMS with transmission electron microscopy to visualize subvesicular compartmentalization to identify the absolute concentration of dopamine in discrete vesicular compartments and whole vesicles.93 This technique has increased the possibility of absolute quantification and direct measurement of the specific contents of nanometer-scale organelles. Figure 3 | Examples of SIMS-based single-cell mass spectrometry. (a) Principle of a SIMS measurement. A primary ion beam sputters the sample surface to generate secondary particles, including neutral particles and ions. The secondary ions are then extracted into the mass spectrometer. Reprinted with permission from ref 88. Copyright 2019 The Royal Society of Chemistry. (b) Schematic of the 3D OrbiSIMS. Reprinted with permission from ref 88. Copyright 2019 The Royal Society of Chemistry. (c) 3D OrbiSIMS mapping of GABA, dopamine, and serotonin in the cornu ammonis region of the mouse hippocampus. Reprinted with permission from ref 94. Copyright 2017 Springer Nature. (d) Schematic of GCIB-SIMS imaging of HeLa cells. Reprinted with permission from ref 96. Copyright 2020 The American Association for the Advancement of Science. Download figure Download PowerPoint Although TOF-SIMS has a high spatial resolution for imaging single cells, it typically lacks the very high mass-resolving power, mass accuracy, and MS/MS capability necessary for the identification of biomolecules. Gilmore and coworkers94 have developed a 3D OrbiSIMS that combines a TOF-SIMS and an Orbitrap mass analyzer to achieve high-speed imaging, good resolution, and excellent sensitivity (Figure 3b). The 3D OrbiSIMS is capable of visualizing exogenous and endogenous metabolites in 3D with subcellular resolution; for example, they imaged the distribution of neurotransmitters—gamma-aminobutyric acid (GABA), dopamine, and serotonin in the mouse hippocampus (Figure 3c). They also putatively identified and mapped the subcellular distribution of 45 glycerophospholipids and 29 sulfoglycosphingolipids and used tandem mass spectrometry to confirm the lipid identities. This technique captured hundreds of metabolites at subcellular resolution in a single measurement resulting in a rich multiplexed data set; thus, it has the potential to shed light on fundamental biological processes. Another limitation of SIMS is that its analysis of biological samples produces evident fragmentation, so molecules are hardly desorbed and measured intact. Gas cluster ion beams (GCIBs) have been used for SIMS-based MSI because they produce less fragmentation than liquid metal ion guns but still provide excellent spatial resolution (<3 μm).89,95 For example, Benkovic and coworkers96 performed low-chemical-damage, in-situ, 3D chemical imaging analyses of intact molecular ions of metabolites using GCIB-SIMS (Figure 3d), such as their analysis of the de novo purine biosynthesis in single HeLa cells. Plasma-based methods The low-temperature plasma (LTP)-based ionization source is highly efficient and easy to operate at a low cost. It is responsive to both polar and nonpolar compounds due to the presence of various reactive species in a plasma, such as electrons, ions, neutral atoms, excited states, radicals, and photons.97–100 The DBDI source, a type of LTP source, has been widely used in the field of mass spectrometry.101–104 Our lab has developed an active capillary plasma ionization based on the DBDI source, which guarantees almost 100% ion transport efficiency due to its direct connection to the inlet capillary of the mass spectrometer.105–113 Since DBDI does not require a vacuum working environment and the associated complex sample preparation process involves a soft ionization process, we have recently applied the DBDI source to the field of single-cell metabolomics.37,38 We developed a high-throughput (∼38 cells/min), label-free DBDI-MS-based single-cell analytical method. Based on the measured metabolites, we were able to distinguish multiple cell types, including embryonic kidney, ductal pancreatic epithelial carcinoma, ductal pancreatic adenocarcinoma, human pancreatic epithelial (H6c7), HeLa, and immortalized brown adopocyte cells (Figure 4a). In addition, we observed an abnormal lipid metabolism in pancreatic cancer cells by DBDI-MS and then proposed and validated that it was caused by the dysregulation of ATP citrate lyase (ACLY). Therefore, ACLY was proposed to be a potential biomarker of pancreatic cancer.37 Figure 4 | Examples of plasma-based single-cell mass spectrometry. (a) Principal component analysis (PCA) of HEK-293T, PANC-1, CFPAC-1, H6c7, HeLa, and iBAs cells measured by DBDI-MS. Reprinted with permission from ref 37. Copyright 2021 John Wiley & Sons. (b) Schematic of the nanoESI-DBDI ionization source. Reprinted with permission from ref 38. Copyright 2022 American Chemical Society. (c) Mass spectra of onion cells acquired in the ESI (3.5 kV) mode and ESI (3.5 kV)–DBDI (2.6 kV) hybrid mode. Reprinted with permission from ref 38. Copyright 2022 American Chemical Society. Download figure Download PowerPoint Metabolites in cells are diverse and vary in polarity. ESI is a powerful tool for detecting polar metabolites, but it is not easy to ionize nonpolar metabolites in single cells by ESI alone. Without information on nonpolar metabolites, some vital information may be overlooked in the study of single-cell metabolomics. To enhance the coverage of metabolites of different polarities in single cells, our lab reported a hybrid ionization source that combines an ESI source and a DBDI source to improve analyte coverage in single cells, mainly targeting nonpolar metabolites (Figure 4b). In brief, the hybrid ionization source can operate in both ESI mode and ESI-DBDI