Here, we present a protocol to accurately and reliably measure metabolites in rare cell types. Technical improvements, including a modified sheath fluid for cell sorting and the generation of relevant blank samples, enable a comprehensive quantification of metabolites with an input of only 5000 cells per sample.
Cellular function critically depends on metabolism, and the function of the underlying metabolic networks can be studied by measuring small molecule intermediates. However, obtaining accurate and reliable measurements of cellular metabolism, particularly in rare cell types like hematopoietic stem cells, has traditionally required pooling cells from multiple animals. A protocol now enables researchers to measure metabolites in rare cell types using only one mouse per sample while generating multiple replicates for more abundant cell types. This reduces the number of animals that are required for a given project. The protocol presented here involves several key differences over traditional metabolomics protocols, such as using 5 g/L NaCl as a sheath fluid, sorting directly into acetonitrile, and utilizing targeted quantification with rigorous use of internal standards, allowing for more accurate and comprehensive measurements of cellular metabolism. Despite the time required for the isolation of single cells, fluorescent staining, and sorting, the protocol can preserve differences among cell types and drug treatments to a large extent.
Metabolism is an essential biological process that occurs in all living cells. Metabolic processes involve a vast network of biochemical reactions that are tightly regulated and interconnected, allowing cells to produce energy and synthesize essential biomolecules1. To understand the function of metabolic networks, researchers measure the levels of small molecule intermediates within cells. These intermediates serve as important indicators of metabolic activity and can reveal critical insights into cellular function.
Mass spectrometry (MS) is the most popular choice for the specific detection of metabolites in complex samples1,2. Nuclear magnetic resonance (NMR) has advantages in the absolute quantification of compounds and structure elucidation, but MS can often resolve more components in complex mixtures such as biofluids or cell extracts. More often than not, MS is combined with prior separation of the compound by capillary electrophoresis (CE), gas chromatography (GC), or liquid chromatography (LC)3. The choice of separation platform is mostly driven by the range of target metabolites and the type of sample and, in a real-world setting, by the availability of machines and expertise. All three separation platforms have a broad and overlapping range of suitable metabolites but different limitations. Briefly, CE can only separate charged molecules and requires a lot of expertise to implement robust analysis of a large number of samples4. GC is limited to molecules that are small and apolar enough to evaporate before decomposing3. Considering all commercially available LC columns, any two metabolites can be separated by this technology5. However, many LC methods exhibit less resolving power than CE or GC methods of similar length.
The typical amount of starting material for metabolomics measurements is usually in the range of 5 x 105 to 5 x 107 cells per sample, 5-50 mg of wet tissue, or 5-50 µL of body fluid6. However, it can be challenging to obtain such amounts of starting material when working with primary cells of rare cell types, such as for example hematopoietic stem cells (HSCs) or circulating tumor cells. These cells are often present in very low numbers and cannot be cultivated without compromising critical cellular features.
HSCs and multipotent progenitor cells (MPPs) are the least differentiated cells of the hematopoietic system and continuously produce new blood cells throughout an organism's life. The regulation of hematopoiesis is of clinical relevance in conditions such as leukemia and anemia. Despite their importance, HSCs and MPPs are among the rarest cells within the hematopoietic system. From a single mouse, typically, about 5000 HSCs can be isolated7,8,9. As traditional metabolomics methods require more input material, pooling cells from multiple mice was often necessary to analyze rare cell types10,11.
Here, we aimed to develop a protocol that enables the measurement of metabolites in as little as 5000 cells per sample to enable the generation of metabolomics data from the HSCs of a single mouse12. At the same time, this method allows to generate multiple replicates from a single mouse for more abundant cell types like lymphocytes. This approach reduces the number of animals required for a given project, thus contributing to the "3R" (reduction, replacement, refinement) of animal experiments.
Metabolites in cells can have very high turnover rates, often in the order of seconds13. However, preparing samples for fluorescence-activated cell sorting (FACS) can take hours, and FACS sorting itself can take minutes to hours, leading to potential alterations in the metabolome due to non-physiological conditions. Some of the reagents used in this protocol (such as ammonium-chloride-potassium [ACK] lysis buffer) can have similar effects. These conditions can cause cellular stress and impact the levels and ratios of metabolites within cells, leading to inaccurate or biased measurements of cellular metabolism14,15,16. The metabolic changes due to sample preparation are sometimes referred to as sorting artifacts. Long digestion protocols and harsh reagents that might be required to produce single-cell suspensions from hard or tough tissues can aggravate this issue. What changes can occur likely depends on the cell type and the processing condition. The precise nature of the changes remains unknown, as the metabolic state of the undisturbed cells in the living tissue cannot be measured.
The protocol presented here involves several key differences compared to traditional methods, namely the use of 5 g/L NaCl as a sheath fluid, sorting directly into extraction buffer, injecting large sample volumes on hydrophilic interaction liquid chromatography-mass spectrometry (HILIC-MS), and utilizing targeted quantification, rigorous use of internal standards and background controls (Figure 1). This protocol has the potential to preserve differences among cell types and between drug treatment and vehicle control to a large extent12. Even for cultured cells, it compares favorably to alternative approaches, such as the more established centrifugation and manual removal of supernatant. However, as sorting artifacts may still occur, data must be interpreted with caution. Despite this limitation, the protocol represents a significant improvement in the field of metabolic profiling, allowing for more accurate and comprehensive measurements of cellular metabolism in rare primary cells12.
The ability to robustly measure broad metabolic profiles in rare primary cells opens the door to new experiments in biomedical research involving these cells. For example, metabolically mediated regulation in HSCs has been shown to impact dormancy self-renewal capacity, with implications for anemia and leukemia11,17. In patient-derived circulating tumor cells, differences in the expression of metabolic genes between tumor and adjacent cells have been shown18,19. This protocol now allows researchers to study these differences systematically on a metabolic level, which is generally regarded as closer to the cellular phenotype than gene expression.
Breeding and husbandry of all mice used for this protocol were conducted in a conventional animal facility at the Max Planck Institute for Immunobiology and Epigenetics (MPI-IE) according to the regulations of the local authorities (Regierungspräsidium Freiburg). Mice were euthanized with CO2 and cervical dislocation by FELASA B-trained personnel following guidelines and regulations approved by the animal welfare committee of the MPI-IE and the local authorities. No animal experimentation was performed, and mice were without health burdens.
NOTE: Highly sensitive analytical methods such as the LC-MS method used in this protocol have the potential to detect even very minor contaminations in the sample. Therefore, it is of utmost importance to minimize such contaminations. The single most important measure is to wear clean gloves whenever touching anything that comes in contact with the samples. This includes for example, preparation of buffers and sheath fluid, labeling of sample tubes, and cleaning laboratory equipment. Moreover, single-use labware should be used whenever possible. Glassware should be rinsed with LC-MS grade solvents prior to use. A clean working environment further helps reduce contaminations.
1. General preparations
2. Isolation of B and T cells from the spleen
3. Isolation of HSCs and MPPs from bone marrow
4. FACS Sorting
5. LC-QQQ-MS measurement
6. Data pre-processing in automRm
NOTE: The following steps discuss conversion from .d to .mzML in msconvert, preparation of metabdb, loading of data and models and metabdb, and general strategies for manual peak review.
FACS sorting enables the isolation of clean populations of different cell types from the same cell suspension (Figure 2 and Figure 3). The specificity of this method relies on the staining of the different cell types with specific surface markers (for example, B cells and T cells from the spleen) or specific combinations of surface markers (for example HSCs and MPPs). Staining of intracellular markers typically requires permeabilization of the cell membrane. This can cause leakage of metabolites, and therefore this type of staining is not suitable to be used in this protocol.
LC provides the separation of metabolites prior to detection by MS (Figure 4). Metabolites elute from the HILIC column approximately sorted by polarity, with less polar compounds eluting early and more polar metabolites eluting later. The signal intensity is determined by the amount of a target metabolite in the sample and a metabolite-specific response factor. Consequently, signal intensities cannot be meaningfully compared across metabolites but only for one metabolite across samples. The 3 peaks that are highlighted in Figure 4 represent 1: niacinamide is detected in both debris and cell extract, albeit at different levels, 2: acetyl-carnitine that is almost exclusively detected in cell extracts, and 3: the internal standard aminoterephatalic acid that is detected at the same level in debris samples and cell extracts.
Metabolic differences among cell types from the same tissue are preserved using the combined FACS-LC-MS protocol (Figure 5). For HSCs and MPPs, cells from 6 mice were required to generate 6 samples, leading to a larger variability within each group compared to B cells and T cells, which were sorted from the same murine spleen. Debris samples representing process blank control samples are clearly separated from samples containing cell extracts. Within the debris samples, a clear separation of the two experiments is visible, highlighting the difference in metabolic environment that the cells experience prior to the sort. When represented in a clustered heatmap (Figure 6), additional information becomes visible, such as the relative levels of metabolites across samples.
Figure 1: Overview of key steps of the presented protocol. This figure has been reprinted with permission from Schoenberger et al.12. Please click here to view a larger version of this figure.
Figure 2: Isolation of pure populations of B cells and T cells from the spleen. Cells and debris events were selected based on forward scatter (FSC) and sideward scatter (SSC) signals. B cells and T cells were selected based on cell type-specific staining signals. This figure has been adapted with permission from Schoenberger et al.12. Please click here to view a larger version of this figure.
Figure 3: Isolation of HSC and MPP from bone marrow. Cells and debris events were selected based on forward scatter (FSC) and sideward scatter (SSC) signals. HSC and MPP populations were selected based on multiple stages of cell type-specific staining signals. This figure has been adapted with permission from Schoenberger et al.12. Please click here to view a larger version of this figure.
Figure 4: Example chromatograms. Samples with (A) 5000 HSCs and (B) 5000 debris events from the same sort. Note that axis share the same scale in both panels. Highlighted metabolites are 1: nicotinamide, 2: acetyl-carnitine, 3: aminoterephtalic acid (ATA). Please click here to view a larger version of this figure.
Figure 5: Principal component analysis (PCA) of metabolic profiles of different cell types and debris background samples from the same experiments. Major separation is observed between the different organs and debris background samples from any organ. In addition, cell types are clearly separated within the cells of each organ. Please click here to view a larger version of this figure.
Figure 6: Heatmap representing metabolites measured across different cell types and matching background control samples (debris). Missing values were imputed with the half-minimum value of the same metabolite across all samples. For plotting, data was normalized to the maximum separately in every row. Please click here to view a larger version of this figure.
Supplementary Table 1: The table lists the compound-specific settings for the selected metabolites, including retention time (RT), polarity (Pol), Precursor m/z (Q1), Fragment m/z (Q3), and collision energy (CE). Please click here to download this File.
The most critical steps for successful implementation of targeted metabolomics using this protocol are 1) a robust staining and gating strategy that will yield clean cell populations 2) precise handling of liquid volumes, 3) reproducible timing of all experimental steps, in particular all steps prior to metabolite extraction. Ideally, all samples belonging to one experiment should be processed and measured in one batch to minimize batch effects22. For larger experiments, we suggest collecting cell extracts at -80 °C and then measuring all extracts in one batch.
When working with low-input samples, clean handling of all materials that are used in the course of the experiments must be emphasized23. Common contaminations are residual buffer, detergents, and skin-care products. The most important measure to minimize contamination is the use of suitable gloves throughout the experiment.
To minimize potential alterations in metabolite composition due to exposure of cells to unphysiological conditions, it is crucial to prepare the samples on ice and with cold reagents (4 °C) is crucial. Also, being as fast as possible before the FACS step will increase the quality of measurement2. The indicated times for antibody staining are as short as reasonable for good staining efficiency. An additional step to block the unspecific binding of antibodies to Fc receptors using Fc-block antibodies could be added before the FACS staining procedure24,25. This prevents contamination of the target cell populations and is especially advisable when working with cell suspensions that include many Fc-receptor high cell types (like monocytes or macrophages) or cells that were cultured without serum. However, it will increase the overall handling time by 10-15 min and therefore may add additional non-physiological variability to the results.
With the use of dynamic MRM methods on modern mass spectrometers, all metabolites listed in compound-specific settings can be recorded with a single injection. Limiting the number of target metabolites allows for the implementation of the described protocol on older mass spectrometers and accelerates data pre-processing and data analysis.
Some level of background signal in the LC-MS data is unavoidable. Therefore, great care must be taken to identify those metabolites that can be detected above the background. Sorting events that are too small to represent intact cells (debris) generates relevant process blank samples that represent matrix-dependent background signal intensities. In addition, internal standards allow the identification of problematic samples that should be excluded from further data analysis26. Both quality assurance measures work best if a sufficiently large number of replicates is available. We recommend using at least six replicates per group.
The authors have nothing to disclose.
The authors would like to thank the animal facility of the Max Planck Institute of Immunobiology and Epigenetics for providing the animals used in this study.
13C yeast extract | Isotopic Solutions | ISO-1 | |
40 µm cell strainer | Corning | 352340 | |
Acetonitrile, LC-MS grade | VWR | 83640.32 | |
ACK lysis buffer | Gibco | 104921 | Alternatively: Lonza, Cat# BP10-548E |
Adenosine diphosphate (ADP) | Sigma Aldrich | A2754 | |
Adenosine monophosphate (AMP) | Sigma Aldrich | A1752 | |
Adenosine triphosphate (ATP) | Sigma Aldrich | A2383 | |
Ammonium Carbonate, HPLC grade | Fisher Scientific | A/3686/50 | |
Atlantis Premier BEH Z-HILIC column (100 x 2.1 mm, 1.7 µm) | Waters | 186009982 | |
B220-A647 | Invitrogen | 103226 | |
B220-PE/Cy7 | BioLegend | 103222 | RRID:AB_313005 |
CD11b-PE/Cy7 | BioLegend | 101216 | RRID:AB_312799 |
CD150-BV605 | BioLegend | 115927 | RRID:AB_11204248 |
CD3-PE | Invitrogen | 12-0031-83 | |
CD48-BV421 | BioLegend | 103428 | RRID:AB_2650894 |
CD4-PE/Cy7 | BioLegend | 100422 | RRID:AB_2660860 |
CD8a-PE/Cy7 | BioLegend | 100722 | RRID:AB_312761 |
cKit-PE | BioLegend | 105808 | RRID:AB_313217 |
Dynabeads Untouched Mouse CD4 Cells Kit | Invitrogen | 11415D | |
FACSAria III | BD | ||
Gr1-PE/Cy7 | BioLegend | 108416 | RRID:AB_313381 |
Heat sealing foil | Neolab | Jul-18 | |
Isoleucine | Sigma Aldrich | 58880 | |
JetStream ESI Source | Agilent | G1958B | |
Leucine | Sigma Aldrich | L8000 | |
Medronic acid | Sigma Aldrich | M9508-1G | |
Methanol, LC-MS grade | Carl Roth | HN41.2 | |
NaCl | Fluka | 31434-1KG | |
PBS | Sigma Aldrich | D8537 | |
Sca1-APC/Cy7 | BioLegend | 108126 | RRID:AB_10645327 |
TER119-PE/Cy7 | BioLegend | 116221 | RRID:AB_2137789 |
Triple Quadrupole Mass Spectrometer | Agilent | 6495B | |
Twin.tec PCR plate 96 well LoBind skirted | Eppendorf | 30129512 | |
UHPLC Autosampler | Agilent | G7157B | |
UHPLC Column Thermostat | Agilent | G7116B | |
UHPLC Pump | Agilent | G7120A | |
UHPLC Sample Thermostat | Agilent | G4761A |