We present a two-step protocol for high-quality mitochondria isolation that is compatible with protein discovery and quantification at a proteome scale. Our protocol does not require genetic engineering and is thus suitable for studying mitochondria from any primary cells and tissues.
Most physiological and disease processes, from central metabolism to immune response to neurodegeneration, involve mitochondria. The mitochondrial proteome is composed of more than 1,000 proteins, and the abundance of each can vary dynamically in response to external stimuli or during disease progression. Here, we describe a protocol for isolating high-quality mitochondria from primary cells and tissues. The two-step procedure comprises (1) mechanical homogenization and differential centrifugation to isolate crude mitochondria, and (2) tag-free immune capture of mitochondria to isolate pure organelles and eliminate contaminants. Mitochondrial proteins from each purification stage are analyzed by quantitative mass spectrometry, and enrichment yields are calculated, allowing the discovery of novel mitochondrial proteins by subtractive proteomics. Our protocol provides a sensitive and comprehensive approach to studying mitochondrial content in cell lines, primary cells, and tissues.
Mitochondria are complex and dynamic organelles able to sense and adapt to the metabolic needs of the cell. Central to the complexity of cellular metabolism, mitochondria act as metabolic hubs where carbohydrate, protein, lipid, nucleic acid, and co-factor metabolism reactions converge1. They also serve as signaling organelles for pathways of the innate immune response and in response to changes in ions and reactive oxygen species2,3. To date, around 1,100 proteins have been mapped to mitochondria4,5,6, yet we can assume that many more remain to be discovered, especially those expressed only in certain cell types or transiently under specific environmental conditions. Developing new approaches for quantifying changes in mitochondrial composition in metabolic states of interest will increase our knowledge of these organelles and highlight novel therapeutic avenues for the disorders characterized by mitochondrial dysfunction7.
Currently, different mitochondria isolation protocols are available, with differing yields and levels of purity8. Centrifugation-based approaches are the most popular, due to their simplicity and low cost. Although suitable for most applications, differential centrifugation has the disadvantage of obtaining lower mitochondrial purity and requiring large amounts of starting material when more complex density gradient-based applications are used. In recent years, new methods for mitochondria isolation have emerged, such as tag-based immune capture ("MITO-IP")9 and fluorescence-activated organelle sorting10. Although both procedures can generate samples with high purity, the former requires genetic engineering to tag mitochondria for affinity purification, making the protocols incompatible with primary material from unmodified organisms or human donors. Meanwhile, the latter is dependent on access to flow cytometry and sorting instruments. Combining different isolation methods offers the promise of generating more robust protocols and increased purity.
Here, we present a new protocol for mitochondria isolation based on the combination of two existing methods: (1) differential centrifugation to isolate a crude mitochondrial fraction, and (2) tag-free immune capture of mitochondria with superparamagnetic beads covalently bound to antibodies against translocase of outer mitochondrial membrane 22 (Tomm22)11, a ubiquitous mitochondrial outer-membrane protein (Figure 1). The procedure we describe is compatible with quantitative protein mass spectrometry, and because it is tag-free and does not require genetic manipulation, it can be applied to a wide range of research models, from cell lines to body fluids to whole animal tissues. Furthermore, the use of two steps in the protocol enables the use of subtractive proteomics6,12 for the discovery of novel mitochondrial proteins and the study of their expression.
Gloves must be worn at all times and cell culture steps performed under a laminar flow hood. The cells are maintained in a 37 °C incubator with 5% CO2. The research presented in this protocol was approved and performed in compliance with the University of Lausanne and Swiss guidelines for the use of animals.
1. Culture of RAW264.7 macrophage cell line
2. Isolation and culture of bone marrow-derived macrophages (BMDMs)
NOTE: The protocol described here is for a single mouse and can be scaled up for multiple mice. Detailed protocols for BMDM isolation and culture have been described elsewhere13,14.
3. Preparation of a crude mitochondrial fraction by differential centrifugation
NOTE: Perform all centrifugation steps at 4 °C. Two centrifuges are required, one with a swing-out rotor and adaptors for conical tubes with a relative centrifugal force of at least 300 x g, the other with a relative centrifugal force of at least 21,000 x g suitable for 1.5 mL tubes. When using adherent cells, use a cell scraper.
4. Superparamagnetic antibody-based purification of mitochondria
NOTE: Perform all the following steps in a cold room at 4 °C.
Three samples with increasing degrees of mitochondrial purity are generated in the present protocol: total cells, crude mitochondria ("mito-crude"), and pure mitochondria ("mito-pure") (Figure 1). We validated the purification of mitochondria from the RAW264.7 macrophage cell line by loading equal protein amounts of each fraction on a gel and immunoblotting, and found that the mitochondrial citrate synthase (Cs) was enriched at each purification step; meanwhile, proteins from the cytosol (GAPDH), the plasma membrane (Na/K ATPase), the nucleus (Lamin B), the lysosomes (Lamp1), and the endoplasmic reticulum (ER) (Pdi) progressively disappeared (Figure 2A). Similar results were obtained using BMDMs. For further validation of the purity and integrity of the isolated mitochondria, electron microscopy on the pure mitochondrial fraction was performed. We observed mitochondria with a classical oval shape and intact cristae surrounded by electron-dense particles corresponding to the antibody-coated beads (Figure 2B). Therefore, it can be concluded that our protocol enriches mitochondria, depletes other cellular components, and maintains mitochondrial structural integrity.
Next, a proteome analysis of each fraction using liquid chromatography coupled to mass spectrometry (LC/MS) was performed. A total of 6,248 proteins in the extract from total cells, 907 of which were previously annotated as mitochondrial in the MitoCarta3.0 inventory5, were identified. After filtering for proteins with a threshold of at least two unique peptides, we calculated an enrichment score for each protein in each sample based on their intensity compared to total cells. We then allocated the proteins to seven major subcellular compartments: mitochondria, ER, lysosomes, Golgi apparatus, cytoskeleton, nucleus, and cytosol, using Gene Ontology (GO)16,17 and MitoCarta3.05 as references. Importantly, an average enrichment for mitochondrial proteins of more than 10-fold and more than 20-fold in the crude and pure mitochondria fractions, respectively, was observed (Figure 2C). In contrast, components of the other six cellular compartments analyzed were depleted during the purification procedure. Of particular note, in the crude mitochondria fraction, we observed a transient enrichment for ER and lysosomal proteins, two classes of contaminant proteins frequently present following differential centrifugation protocols18. This was possibly due to organelle-organelle interactions and similar coefficients of sedimentation, especially for lysosomes, which are highly abundant in macrophages19. While both were mostly depleted after immune capture, we detected a small signal for proteins from the ER-mitochondria contact sites in the mito-pure fraction.
We then directly compared the protein abundance from the total cells and from mito-pure samples and observed two distinct populations, corresponding to mitochondrial and non-mitochondrial proteins (Figure 2D). While the vast majority of MitoCarta proteins clustered together, we found a few (<5%) that clustered with non-MitoCarta proteins. These proteins may represent (1) cytosolic mitochondria-interacting proteins (a novel category annotated in version 3.0 of MitoCarta), (2) dual-localized proteins, or (3) mis-annotated proteins. Conversely, we found a few instances of non-MitoCarta proteins clustering with mitochondrial proteins. While such proteins may represent contaminants of the isolation procedure, they may also represent proteins not previously classified as being present in mitochondria.
To investigate this new class of potential mitochondrial proteins, subtractive proteomics, an approach that has proved useful for the discovery of organellar proteomes, including mitochondria6,12, was used. Subtractive proteomics assumes that mitochondria should become enriched during the purification steps, and contaminants should become depleted6. For example, whereas contaminants may accumulate during differential centrifugation (e.g., due to similar sedimentation properties) or during immune capture (e.g., due to non-specific antibody binding), only bona fide mitochondrial proteins should significantly accumulate in both. It is thus possible to filter out proteins that were found in the pure mitochondria fraction but showed inconsistent patterns of enrichment. In the present example with RAW264.7 cells, by setting a threshold for unique peptides of ≥1 for the mito-crude and mito-pure samples, and using stringent thresholds of enrichment, we were able to refine the list of recovered mitochondrial proteomes from 1,127 proteins initially found in the crude mitochondrial fraction after differential centrifugation, down to 481 proteins following the second round of purification using Tomm22 immunoselection. The reduced number of MitoCarta annotated proteins in the mito-pure fraction reflects the high stringency applied for selection. Interestingly, 70 of the proteins present in the mito-pure fraction were not present in the MitoCarta3.0 inventory (Figure 3A, B). These latter proteins may represent potential novel mitochondrial candidate proteins, which may only be expressed in the RAW264.7 macrophage cell line and in related cells, and which merit further investigation.
Figure 1: Illustration of the two-step, tag-free mitochondria isolation protocol. (A) A cell suspension is disrupted through a 25 G needle. (B) Nuclei and whole cells are separated by centrifugation at 2,000 x g and the supernatant is saved. (C) Crude mitochondria are isolated by differential centrifugation of the supernatant at 13,000 x g (mito-crude). (D) Crude mitochondria are then incubated with Tomm22 antibodies (Ab) covalently linked to superparamagnetic beads. (E) The mitochondria-Tomm22 antibody-beads complexes are separated from contaminants using magnetic columns and eluted. (F) Pure mitochondria are collected and concentrated by centrifugation (mito-pure). Please click here to view a larger version of this figure.
Figure 2. Representative results of mitochondria isolation from two macrophage sources. (A) Protein immunoblot analysis of RAW264.7 (top) and BMDM cells (bottom) using antibodies to mitochondrial citrate synthase (Cs – mitochondria), glyceraldehyde 3-phosphate dehydrogenase (Gapdh – cytosol), sodium-potassium pump (Na/K ATPase – plasma membrane), Lamin B (Lamin B – nucleus), lysosomal-associated membrane protein 1 (Lamp1 – lysosome), and protein disulfide-isomerase (Pdi – ER). (B) Electron microscopy of purified mitochondria from RAW264.7 cells. High density particles surrounding mitochondria correspond to the Tomm22 beads that are carried on with mito-pure samples after elution from the columns. Scale bars: 80 nm. (C) Enrichment scores across total cells, mito-crude, and mito-pure from seven cellular compartments in RAW264.7 cells. MitoCarta3.0 and GO were used for protein annotation and the average scores are represented. Abbreviation: ER = endoplasmic reticulum. (D) Protein abundance values (riBAQ) for proteins in total cells and mito-pure samples from RAW264.7 cells. MitoCarta3.0 proteins are shown in orange. Please click here to view a larger version of this figure.
Figure 3. Discovery of novel mitochondrial proteins using subtractive proteomics. (A) Subtractive proteomics strategy for the discovery of novel mitochondrial proteins. High selection thresholds (4x and 2x) are applied to minimize the selection of false positives. (B) Enrichment yields (fold of total cells) of new mitochondrial candidate proteins not previously annotated in the MitoCarta3.0 inventory. Please click here to view a larger version of this figure.
We have combined differential centrifugation and immunocapture to achieve an improved purity for mitochondria isolation. Our procedure allows access to primary material for the identification and characterization of novel mitochondrial proteins. The protocol is straightforward and robust, and can be applied to cell lines, primary cells, and tissues without the need for genetic modification. We have validated our protocol by immunoblotting and proteomics analyses on samples taken at different stages throughout the purification procedure.
In comparison to single isolation methods, the combination of enrichment steps of different natures – here, centrifugation and immune labelling – generates a more robust protocol to isolate mitochondria. This is because, while mitochondrial proteins will become enriched in both purifications, it is unlikely that contaminants will also be enriched after both enrichment steps. Although high mitochondrial purity can also be achieved by density gradient ultracentrifugation, this approach requires a large amount of starting material and access to an ultracentrifuge. Lastly, in contrast to recent methods based on tag-based mitochondrial isolation20, our approach does not require genetic modification of the sample, making it suitable for primary material from any source.
Some technical and biological considerations need to be taken into account in the experimental design when applying our protocol. (1) The amount of starting material is critical in order to obtain sufficient material. Inevitably, a small number of mitochondria will be lost during homogenization (step 3.10), as not all cells are lysed, or during the three column washes (step 4.6). While our protocol focuses on purity over yield, the efficiency of mitochondria isolation, and hence their yield, has not been measured or optimized. Using more Tomm22 beads and more columns is expected to increase the yield of mitochondria recovery. At the same time, a thorough optimization of the homogenization step can also lead to improved mitochondrial yield. This protocol and the initial cell numbers we report here for RAW264.7 cells and BMDMs are adequate for proteomics and can be adjusted for other applications. In the case of primary BMDMs, we found that a single mouse was sufficient for one replicate. When necessary, the procedure can be scaled up to isolate BMDMs from multiple animals, which can then be pooled to obtain sufficient material. The cell number can be optimized depending on the cell type, its size, and its mitochondrial content. (2) Tomm22 is expressed on mitochondria from all cell types and tissues21, but its expression level may vary. Therefore, when designing an experiment to compare different conditions, it is important to ensure that the expression levels of Tomm22 are comparable. Furthermore, due to the ubiquitous expression of Tomm22, it is not possible to study cell-type specific mitochondrial proteins within complex tissues. (3) The time necessary to generate pure mitochondria (around 2.5 h) is incompatible with a study of transient events, such as changes in metabolic profiles. In this case, we recommend direct tag-based immune capture9, which also allows studying cell-type specific mitochondria in vivo20. (4) Although studies on isolated mitochondria obtained using Tomm22 antibody-labeled beads alone have shown activity in functional assays11, it remains to be determined whether mitochondria generated with our protocol are compatible with downstream activity-based assays. MitoTracker or tetramethylrhodamine methyl ester perchlorate (TMRM) staining, or respirometry measurements, are potential approaches to quantifying the functionality of isolated mitochondria22. (5) After eluting the "mito-pure" sample from the column, some Tomm22 beads will be present in the pure mitochondria fraction (Figure 2B). While we have observed no interference with trypsin digestion and protein mass spectrometry, the presence of these beads and the immunoglobulins should be taken into consideration in other downstream applications. The Tomm22 antibody is a monoclonal antibody produced in mice23, and therefore it is important to keep in mind that, when using secondary antibodies against mice in immunoblotting, it will generate unspecific bands at the size of the immunoglobulin chains. (6) Complete homogenization of the cell suspension is key to the successful isolation of mitochondria. Here, we use a syringe with a 25 G needle to lyse both RAW264.7 cells and BMDMs. However, depending on the cell type and its size, other mechanical homogenization methods, such as use of a Dounce homogenizer, or more controlled approaches like cell homogenizer devices, may be more suitable. Non-mechanical homogenization methods, such as gentle sonication, can also be considered. Tissue homogenization approaches are further discussed in other studies24,25. (7) Although validation by immunoblotting is the most straightforward and cheaper method, its results might not always correlate with changes at the whole organelle level. That is why we recommend using proteomics to fully validate the enrichment or depletion of mitochondria and other organelles, respectively.
The two-step mitochondria purification protocol described here has allowed us to generate sequential samples with increasing mitochondrial purity, and this has enabled us to discover novel mitochondrial protein candidates through subtractive proteomics12. For our analysis, we use stringent thresholds to select for significantly enriched mitochondrial proteins, and although this may fail to identify some known mitochondrial proteins (Figure 3A), the false-positive rate for new mitochondrial protein discovery is decreased. Nevertheless, it is important to stress that any candidate proteins revealed by our protocol must be validated through orthogonal approaches. We recommend carboxy-terminal GFP-tagging or the use of antibodies against the endogenous protein to validate association with mitochondria either microscopically or by protease protection assays.
The direct application of our method in the case of unmodified cells and tissues offers a powerful tool to investigate how mitochondria change and adapt to their environment in healthy and disease conditions. Application of our protocol to cell lines, animal disease models, human fluids, and even biopsies from surgery may prove to be particularly useful to enhance our understanding of mitochondria and their associated disorders.
The authors have nothing to disclose.
We thank Manfredo Quadroni, the Protein Analysis Facility, and the Electron Microscopy Facility at the University of Lausanne for their help. We also thank H.G. Sprenger, K. Maundrell, and members of the Jourdain laboratory for advice and feedback on the manuscript. This work was supported by the Foundation Pierre-Mercier pour la Science, and the Swiss National Science Foundation (project grant 310030_200796).
1 mL syringe | BD Plastipal | 309628 | |
25 G Needle | BD Microlance | 300400 | |
40 µm cell strainer | Corning | 352340 | |
Anti-TOM22 Microbeads, mouse | Miltenyi Biotec | 130-127-693 | |
Cell scraper | FisherScientific | 11577692 | |
DMEM, high glucose, GlutaMAX | ThermoFisher | 31966 | |
Ethylenediaminetetraacetic acid | FisherScientific | BP-120-1 | |
Fetal bovine serum | Gibco | 10270 | |
HEPES | BioConcept | 5-31F00-H | |
LS columns and plungers | Miltenyi Biotec | 130-042-401 | |
Macrophage colony-stimulating factor | Immunotools | 12343115 | |
Mannitol | Sigma | M4125 | |
Sodium chloride | Sigma | 71380 | |
Sucrose | Sigma | S1888 | |
Penicillin/Streptomycin | BioConcept | 4-01F00-H | |
Petri dishes | Corning | BH93B-102 | |
Phosphate-buffered saline 10X | Eurobio Scientific | CS3PBS01-01 | |
QuadroMACS Separator | Miltenyi Biotec | 130-090-976 | |
Vi-CELL BLU Cell Viability Analyzer | Beckman Coulter | C19196 |