Research on mitochondrial protein–protein interaction (PPI) networks is organelle-constrained work. The first decisions are which mitochondrial compartment the bait occupies, how mitochondria will be isolated, and which capture chemistry can see membrane or matrix contacts. Mitochondria produce ATP, but they also participate in apoptosis, metabolic control, innate-immune signaling, and oxidative-stress responses. Those jobs are carried by complexes and dynamic neighborhoods, not by a single subunit abundance value. High-resolution mass spectrometry and systems analysis have moved many projects from one pair to a bait-centered table. They have not made an endoplasmic-reticulum-contaminated pellet into a mitochondrial interactome.

Human mitochondria are catalogued at about 1,100 proteins in resources such as MitoCarta. That inventory is a starting proteome, not a count of proven interactions. Oxidative phosphorylation (OXPHOS), the tricarboxylic acid (TCA) cycle, fatty-acid β-oxidation, fusion and fission, and mitophagy are the usual pathway frames. Most of those proteins act in assemblies that cross the outer membrane (OMM), intermembrane space (IMS), inner membrane (IMM), or matrix. A network that ignores compartment and contamination will mix true mitochondrial neighbors with post-lysis or microsomal artifacts.

Researchers who already have a mitochondrial bait, an isolation plan, or a project objective can also review the MtoZ Biolabs Protein-Protein Interaction Analysis Service for project-specific feasibility and analysis planning.

Why Mitochondrial PPI Networks Are Studied

A mitochondrial PPI study can nominate new modules, connect metabolic pathways, and rank disease-associated proteins for follow-up in neurodegeneration, metabolic disease, or cancer models. Those uses justify the experiment. They do not produce a system-level model or a therapeutic target from one AP-MS table. Dual-localized proteins and ER–mitochondria contact sites are real biology. They are also the main source of over-read edges if purity is not measured.

Figure 1. About 1,100 catalogued human mitochondrial proteins sit in OMM, IMS, IMM, or matrix pathways. The compartment is named before the capture method.

Name the organelle fraction and the compartment first. Throughput language does not correct ER contamination.

Overall Workflow

A usable mitochondrial PPI campaign usually follows this sequence:

Sample preparation.

Mitochondrial isolation and a purity check.

Interaction capture.

LC-MS/MS identification.

Quantitative comparison and scoring.

Network construction with experimental edges kept separate from database overlays.

High-resolution mass spectrometry is the common readout after capture. It is not the step that creates specificity.

Figure 2. Isolation purity and matched controls decide which proteins may enter the network. The instrument name does not.

Mitochondrial Isolation and Purification

A defined mitochondrial fraction is a prerequisite. Isolation can also break weak contacts or mix compartments after lysis, so purity and speed both matter.

1. Differential centrifugation

Cells or tissue are lysed, nuclei are removed at low speed, mitochondria are pelleted at medium speed, and the pellet is washed. The procedure is simple and scales. The pellet commonly retains ER and other membranes. Those contaminants become false mitochondrial partners if they are not flagged.

2. Density-gradient centrifugation

Percoll or sucrose gradients improve mitochondrial purity and are usually the better input for interaction work. High purity reduces some false positives. It does not restore interactions that were lost during homogenization, and it does not prove that every protein in the fraction is mitochondrial.

Marker immunoblots or proteomic contamination indices for ER, cytosol, and lysosome should be reported with the fraction. Without that check, later SAINT filtering is asked to do isolation’s job.

Interaction Capture Techniques

1. Co-immunoprecipitation (Co-IP)

An antibody captures a bait and proteins that survive wash. The eluate can be read by Western or by MS. Co-IP fits a nominated mitochondrial pair or a small complex when an IP-grade antibody exists. It is not automatically highly specific. Transient IMM contacts and detergent-sensitive OXPHOS neighbors are often lost. For pairwise endogenous rules, How Can Endogenous Interactions Between Two Proteins Be Verified? provides a complementary discussion of native versus tagged claims.

2. Affinity purification–mass spectrometry (AP-MS)

FLAG, HA, or similar tags on a mitochondrial bait allow systematic partner inventories and quantitative comparison. The tag can mislocalize a protein that depends on a precise import sequence, so localization of the tagged bait should be checked before the table is interpreted as mitochondrial. Empty-tag and bead-only controls belong in the same isolation batch.

3. Proximity labeling: BioID, TurboID, and APEX

A ligase or peroxidase fused to a compartment-targeted bait biotinylates nearby proteins in living cells. Streptavidin capture and MS then identify those proteins. The class is often more practical than wash-based Co-IP for some mitochondrial membrane neighborhoods. The table reports proximity inside a labeling radius, not wash-stable binding. Targeting peptides must place the enzyme in the intended compartment; a matrix claim cannot be written from an OMM fusion.

Figure 3. Co-IP reports wash-stable native or nominated complexes, AP-MS reports a tagged bait inventory, and mito-targeted BioID, TurboID, or APEX reports live-cell proximity.

For AP-MS versus proximity-labeling evidence type, How to Identify Protein–Protein Interactions Using Mass Spectrometry provides a more detailed explanation of experimental limits.

Mass Spectrometry After Capture

Typical processing is trypsin digestion, LC separation, tandem MS, and a database search. Orbitrap-class instruments improve resolution and help recover lower-abundance peptides. Label-free, TMT, or SILAC comparison can rank bait-dependent proteins when isolation and capture stay matched. Statistical filters such as SAINT, fold change, and FDR are tools for one control model. They are not a universal accuracy setting and they cannot remove ER proteins that were physically present in a dirty pellet.

Bioinformatics of Mitochondrial PPI Networks

1. Filtering

Keep proteins that pass the stated score against matched controls. Mitochondrial annotation (for example MitoCarta) can label nodes. It should not be used as the only specificity filter.

2. Network construction

Draw experimental edges first. STRING, BioGRID, and IntAct can overlay prior knowledge. A STRING edge among two mitochondrial proteins is not a contact measured in the isolated fraction. For how overlay graphs should be analyzed, How to Build and Analyze Protein-Protein Interaction Networks? provides a more detailed build-versus-overlay framework.

3. Enrichment and topology

GO, KEGG, and complex annotations interpret a module. Degree and other topology scores rank nodes on this graph. High-degree proteins are not automatically disease targets.

Typical Research Settings

Neurodegenerative models, including Alzheimer- and Parkinson-related systems, often show mitochondrial dysfunction. A PPI table can nominate pairs near quality-control or OXPHOS nodes. It does not demonstrate the disease mechanism.

Cancer metabolic reprogramming can change mitochondrial neighborhoods. Network analysis can rank regulators for follow-up. It does not reveal anticancer targets.

Innate-immune signaling through complexes such as MAVS is a mitochondrial-membrane PPI problem. Systematic capture can recover neighbors of a defined bait. Novel immune mechanisms still need orthogonal genetics or biochemistry.

Experimental Limits and Design Boundaries

Differential pellets mix ER. Harsh detergents dissolve IMM supercomplexes or create new ones. Proximity enzymes label across a radius that can include passing proteins. Import defects from tags create cytosolic bait and cytosolic prey. Diagnostic or target-discovery language sits outside ordinary research mitochondrial PPI reporting unless a separately validated assay framework is in place.

Project Planning and Service Scope

A usable plan names the compartment, the isolation and purity markers, the capture class, and whether STRING will be used only as overlay. MtoZ Biolabs scopes Co-IP, AP-MS, and proximity-labeling proteomics on Orbitrap-class LC-MS/MS, label-free or TMT comparison when occupancy must be compared, and network annotation after the experimental table is closed. Recovery of hydrophobic IMM partners still depends on detergent, targeting, and controls, not on calling the study mitochondrial.

Once the fraction, capture route, and scoring rule are defined, the project can be scoped as a mitochondrial PPI network study. Researchers can review the MtoZ Biolabs Protein-Protein Interaction Analysis Service for project-specific sample evaluation, feasibility assessment, and workflow planning.

Frequently Asked Questions

1. Is a mitochondrial proteome list a PPI network?

No. A proteome says which proteins are in the fraction. A PPI network requires captured edges and a stated evidence type.

2. Why is density-gradient isolation preferred for interaction work?

It usually reduces ER and other membrane contamination that otherwise appears as false mitochondrial partners. It does not restore contacts lost during homogenization.

3. Are BioID, TurboID, or APEX better for mitochondrial membrane proteins?

They are often more practical for some membrane neighborhoods because they do not require the pair to survive a wash. They still report proximity, and the enzyme must sit in the correct compartment.

4. Can SAINT or TMT correct a contaminated mitochondrial pellet?

No. Scoring compares bait and control in the same fraction. Contaminants that co-isolate with both will remain.

5. Does STRING of mitochondrial proteins reconstruct the organelle interactome?

No. It overlays public associations. Sample-level mitochondrial edges still come from Co-IP, AP-MS, or proximity labeling on that fraction.

Reference

1. Rath S., Sharma R., Gupta R., et al. (2021). MitoCarta3.0: an updated mitochondrial proteome now with sub-organelle localization and pathway annotations. Nucleic Acids Res., 49, D1541-D1547. https://doi.org/10.1093/nar/gkaa1011
2. Pagliarini D.J., Calvo S.E., Chang B., et al. (2008). A mitochondrial protein compendium elucidates complex I disease biology. Cell, 134, 112-123. https://doi.org/10.1016/j.cell.2008.06.016
3. Roux K.J., Kim D.I., Raida M., Burke B. (2012). A promiscuous biotin ligase fusion protein identifies proximal and interacting proteins in mammalian cells. J. Cell Biol., 196, 801-810. https://doi.org/10.1083/jcb.201112098
4. Hung V., Zou P., Rhee H.W., et al. (2014). Proteomic mapping of the human mitochondrial intermembrane space in live cells via ratiometric APEX tagging. Mol. Cell, 55, 332-341. https://doi.org/10.1016/j.molcel.2014.06.003
5. Gingras A.C., Gstaiger M., Raught B., Aebersold R. (2007). Analysis of protein complexes using mass spectrometry. Nat. Rev. Mol. Cell Biol., 8, 645-654. https://doi.org/10.1038/nrm2208
6. Szklarczyk D., Gable A.L., Lyon D., et al. (2019). STRING v11: protein-protein association networks with increased coverage, supporting functional discovery in genome-wide experimental datasets. Nucleic Acids Res., 47, D607-D613. https://doi.org/10.1093/nar/gky1131