ELL is in Cajal bodies - or nuclear speckles, depending which paper you read - and the MLL-ELL fusion empties them

UniProt P55199 carries three subnuclear statements at once:

Nucleus                {ECO:0000269|PubMed:11418481, ECO:0000269|PubMed:12446457}
Nucleus speckle        {ECO:0000269|PubMed:11418481, ECO:0000269|PubMed:12446457}
Nucleus, Cajal body    {ECO:0000269|PubMed:12686606, 22195968, 23932780}

Nuclear speckles and Cajal bodies are different organelles. Speckles are SC35/SRSF2-SON-SRRM2 interchromatin granule clusters; Cajal bodies are coilin-positive snRNP-assembly bodies. They are both punctate and they are both nucleoplasmic, and at confocal resolution in 2001 a small number of bright foci looks the same either way.

What actually happened, read across the three papers, is that the same laboratory described the same foci twice under two names: Simone et al. 2001 and 2003 report "a distinct nuclear speckled pattern" for endogenous EAF1/ELL and EAF2/ELL 12; Polak et al. 2003 co-stain with p80 coilin and conclude the foci are Cajal bodies 3. The 2003 Cajal-body assignment is the better-controlled one because it uses a marker; the 2001/2003 'speckle' descriptions used no speckle marker at all.

Nobody in the subsequent literature says this. No paper read for this entry notes that the two names refer to the same observation, and UniProt lists them as co-existing compartments, which a downstream reader will take as two independent localisations. What would settle it: one experiment co-staining endogenous ELL with coilin AND with SON or SRRM2 in the same cells. Three colours, one afternoon, and it has not been done in twenty-three years.

This is the reason the entry matters to this graph. Most KMT2A-fusion localisation claims are inferences. Here there is a measurement:

• Transfection of MLL-ELL into HeLa cells delocalises EAF1 from its focal distribution to a diffuse nucleoplasmic pattern 1.
• In leukaemic cell lines from mice transplanted with MLL-ELL-transduced bone marrow, EAF1 foci are not detected at all 1, and p80 coilin is delocalised from Cajal bodies 3.
• Nuclear/cytoplasmic fractionation shows reduced nuclear p80 coilin and EAF1 in MLL-ELL leukaemia cells 3.
• The structure-function background makes this load-bearing: ELL's elongation domain is dispensable for MLL-ELL immortalisation while its EAF1-interaction domain is necessary and sufficient, and a heterologous MLL-EAF1 fusion immortalises progenitors [E3, introduction].

So the partner contribution that transforms is the one whose normal job is to concentrate ELL in a nuclear body, and expressing the fusion destroys that body. That is a mechanistic story with a subnuclear compartment at its centre, and it is the template for the question this beat keeps hitting for other partners: does the fusion take the partner's compartment with it, or dissolve it? For MLL-ELL the answer is dissolve.

Rule 73 on the systems: HeLa transient transfection (human, non-haematopoietic) and murine MLL-ELL myeloid leukaemia lines. Neither is human infant B-lineage KMT2A::AFF1, and MLL-ELL is an AML fusion.

ELL and EAF1 leave the bodies when Pol II is inhibited with actinomycin D, DRB or alpha-amanitin 3. A compartment that dissolves when transcription stops is a product of transcription as much as a site of it, and this is the standard caveat for any transcription-associated nuclear body.

The function came later: ELL is a subunit of the Little Elongation Complex (LEC), which translocates with ICE1 to Cajal and histone locus bodies at snRNA genes and functions at both initiation and elongation of Pol II-transcribed snRNA genes 4. That gives the Cajal-body localisation a substrate - snRNA gene transcription - and makes the MLL-ELL disruption a prediction about snRNA output that, as far as this beat could find, nobody has tested in leukaemia.

Rule 74. The ELL/EAF1 imaging used in-house reagents: affinity-purified polyclonal anti-ELL antiserum at 1:250 and an anti-EAF1 monoclonal at 1:50, with rabbit 288 anti-p80 coilin (gift of E. Chan) at 1:250 and a human anti-coilin autoantibody (gift of D. Bloch) at 1:200; markers included mouse anti-fibrillarin (Cytoskeleton) 1:1000 and anti-human-nucleoli (Chemicon) 1:2000. None of the ELL or EAF1 reagents is a catalogue clone, so the experiment is not straightforwardly repeatable today - which is itself a reason the speckle/Cajal question has stayed open.

Note also that PMID 12686606 carries a published correction (Mol Biol Cell 2003;14(5)). The PubMed abstract shows a [corrected] marker on the fractionation sentence; the PMC full text renders that sentence in a garbled form ("diminished expression of p80 coilin and EAF1 are not present in the nuclei"). The safe reading, which both versions support, is reduced nuclear p80 coilin and EAF1 in MLL-ELL leukaemia cells.