KAT8/MOF is not only a chromatin enzyme: a measured mitochondrial pool binds mtDNA, and the chromatin address depends on which complex it is in

A nuclear chromatin enzyme reported in mitochondria is the classic false positive - a fractionation artefact or an antibody cross-reaction. Chatterjee et al. built the case that it is not 2:

• MOF and a subset of NSL complex partners are found in mitochondria, not MOF alone;
• MOF binds mtDNA, and that binding is dependent on KANSL3 - a specific molecular requirement, not a bulk association;
• re-supplying the mitochondrial pool of MOF rescues respiratory and mtDNA transcriptional defects caused by MOF loss, while a catalytically deficient mutant does not - a rescue with an activity control;
• conditional Mof knockout in mouse heart produces hypertrophic cardiomyopathy, cardiac failure and mitochondrial degeneration.

That combination - a partner requirement, a compartment-specific rescue, an activity control and an organ phenotype - is a much stronger structure than a colocalisation image, and it is why UniProt carries Mitochondrion {ECO:0000269|PubMed:27768893} as an experimental location.

Rule 73: mouse heart and cultured aerobically respiring cells. Nothing here is leukaemia, and the cardiac phenotype is the tissue with the highest oxidative demand rather than a generic readout.

UniProt states two different genomic distributions for the same protein:

As part of the NSL complex, associates with the proximal part of promoters
  and transcription start sites          {ECO:0000269|PubMed:33657400}
As part of the MSL complex, associates with gene bodies   (By similarity)

Note the asymmetry: the NSL/promoter statement is experimental in human; the MSL/gene-body statement is ECO:0000250 transferred from mouse Q9D1P2. An import that flattens both into 'KAT8: chromatin' loses the distinction and also promotes a by-similarity statement to a measured one.

The distinction matters here because of what the complexes do 4: KAT8 in NSL catalyses H4K5ac and H4K8ac and is essential for cell survival, stimulating transcription initiation at housekeeping promoters; KAT8 in MSL catalyses the bulk of H4K16ac, and MSL proteins and H4K16ac are not required for proliferation or chromatin accessibility. So the same enzyme is essential in one address and dispensable in the other.

This graph already holds hyp-wdr5-partition-nsl-versus-compass and exp-wdr5-pool-exchange-kansl1-degron. The KAT8 data are the cleanest existing demonstration that a shared subunit's location and essentiality partition by complex, which is precisely the shape of that hypothesis for WDR5.

Two importin-alpha-1-specific NLSs sit in the human MOF N-terminus; the crystal structure of the MOF NLSs bound to importin alpha1 shows them occupying the major and minor sites respectively, with NLS2 adopting an unexpected alpha-helical conformation and making contacts beyond the minor site. Mutating key residues on either partner reduces the interaction and reduces nuclear import of MOF 3.

That is a well-controlled import mechanism. It is also, like AF10's NLS, an answer to can it be imported rather than where is it, and it does not speak to the mitochondrial pool at all - the two literatures do not cite each other's compartments.

KAT8 sits at degree 20 in this graph, is a DepMap floor-compression example noted in the standing brief (99.5% of lines depleted), and is the catalytic subunit of the NSL complex that NUP98 fusions engage. And:

• no image or fractionation of KAT8 in any KMT2A-rearranged cell was found;
• nobody has asked whether the mitochondrial pool exists in leukaemic blasts, although leukaemia stem cells are notably dependent on oxidative phosphorylation;
• no measurement distinguishes the NSL-bound and MSL-bound pools in this disease, so a KAT8 dependency here cannot currently be attributed to a complex, and therefore not to a chromatin address.