Proteins located in the same intracellular compartment do not necessarily interact, whereas proteins located in different compartments are not necessarily unable to interact.

Proteins located in different compartments and able to interact may be kept apart that way until particular conditions are reached.

For example, cytochrome c, which is a small protein associated with the inner membrane of the mitochondria and plays a key role in cellular respiration, can be released to the cytoplasm interacting with Apaf-1 and triggering caspase-mediated apoptosis.

In addition, it has been shown that cytochrome c is able to bind many nuclear proteins.

See https://pubmed.ncbi.nlm.nih.gov/31663111/ and many other publications from this research group.

I assume by saying "mitochondrial protein" you mean the protein is encoded by mitochondrial DNA.

Here are some heads-up

1. CoIP indirectly reflects protein-protein interactions. Even if the protein is precipitated in the elution, does not necessarily mean these two proteins are physically interacting. They might be part of a protein complex in which the mitochondrial protein is being post-translational modified. I won't be surprised if the bound proteins are enzymes or scaffold proteins.

2. Protein hits mapped by MS/MS are based on an algorithm, which is a matter of probability. So, the peptides were blasted to the database and locate the matches. So, there is a possibility the identified proteins were matched with low confidence.

3. Yes, proteins can interact with each other in a tube, and this is the foundation for in vitro binding assays.