Cytoplasmic localisation in KMT2A fusions: the SEPTIN case has an image, the GPHN case has none, and neither partner wins the compartment
also called MLL-SEPT6 localisation · MLL-GPHN localisation · cytoplasmic partner compartment
A few of MLL's fusion partners are proteins that normally work outside the nucleus - at cell membranes or on the scaffolding that holds nerve-cell receptors in place. When they are fused to MLL the chimera is pulled into the nucleus anyway, and for the gephyrin fusion nobody has ever taken the picture at all.
The same thing, for a scientist
Two KMT2A partner classes are genuinely cytoplasmic in their own right: the septins and gephyrin (GPHN). The septin case has a direct image - by confocal immunofluorescence, MLL-SEPT6 distributed as small nuclear speckles or dots in 293T cells while overexpressed SEPT6 alone did not shift from its physiological cytoplasmic localisation [E1]. The gephyrin case has none. GPHN entered this literature as a fusion partner in t(11;14)(q23;q24) [E2][E3], and the functional work showed that a **15-amino-acid tubulin-binding domain** of gephyrin is necessary and sufficient to transform haematopoietic progenitors and confers oligomerisation on MLL, with MLL-GPHN producing myeloid and lymphoid leukaemias in mice [E4]. That is an oligomerisation result, not a localisation result, and as of 2026-09-20 the entire literature intersecting MLL/KMT2A with GPHN/gephyrin is **three records**, none of which reports where the fusion protein sits [E5]. So the graph should hold 'MLL-GPHN is nuclear' as an inference from the retained KMT2A import determinant, not as an observation. The organising point for this beat is that in both cases the compartment is set by retained KMT2A sequence and the partner contributes an oligomerisation or scaffolding module rather than an address.
What is actually imaged, and in what system
| fusion | imaged? | system | what was seen | attributable? |
|---|---|---|---|---|
| MLL-SEPT6 | yes | transiently transfected 293T, confocal IF | small nuclear speckles/dots; SEPT6 alone stays cytoplasmic | yes - the fusion is the only species carrying the construct, but overexpressed |
| MLL-GPHN | no | - | - | - |
| MLL-AF6 | yes | not stated in abstract; IF + cell fractionation | nuclear despite cytoplasmic AF6 | partially - construct-based |
The septin entry is the strongest of the three because it carries its own internal control in the same paper: the partner alone, overexpressed to the same degree, does not move. That controls for the commonest overexpression artefact, which is that a protein pushed past its physiological level redistributes.
It does not control for the second artefact, which is that a punctate appearance at high expression is the default for a multivalent chromatin-binding protein. This graph's condensate-standards entry applies in full.
The GPHN gap, stated as a bounded absence
On 2026-09-20, (MLL OR KMT2A) AND (GPHN OR gephyrin) returns 3 records - PMIDs 14751928, 11579461 and 11289145 - with a query translation containing no injected [UID] term 5. All three are accounted for: the original t(11;14) cloning 2, a second independent report of the same fusion in acute monoblastic leukaemia 3, and the oligomerisation-domain functional paper 4. None reports imaging or fractionation of the fusion protein.
The absence is bounded by the symbol set, and the retired-symbol warning that applies to septins applies here too: gephyrin has published as GPH and KIAA1385 in older work, and a search that misses those would undercount. The three records are, however, the complete MLL-intersecting set under both current symbols, and the fusion is rare enough that a large hidden literature is implausible.
The second, harder question nobody has asked for either partner
This graph already records the reverse-traffic mechanism, documented for AFDN: MLL-AF6-rearranged cells relocate endogenous wild-type AF6 from cytoplasm to nucleus, with consequent RAS activation, and silencing the fusion restores cytoplasmic AF6 and reduces RAS-GTP 6. A cytoplasmic partner can therefore contribute twice - once as a module inside the chimera and once by depleting the cytoplasm of its wild-type self.
For gephyrin that second contribution is not a curiosity. Gephyrin's day job is clustering glycine and GABA-A receptors at inhibitory synapses and catalysing molybdenum cofactor synthesis; depleting a cytoplasmic gephyrin pool is a loss-of-function question with its own phenotype. For septins it would mean depleting a cytoskeletal filament component. Neither has been measured, and the measurement is a blot of the cytoplasmic fraction in a rearranged cell against a non-rearranged control - the same fractionation design this beat proposes for the fusion itself.
| gphn literature records | 3 |
|---|---|
| gphn minimal transforming module AA | 15 |
| gphn module | tubulin-binding domain, oligomerisation |
partners with cytoplasmic baseline
[ "gene-septin6", "gene-gphn", "protein-afdn-domains" ]
imaged
{
"MLL-SEPT6": "confocal IF, 293T, small nuclear speckles/dots",
"MLL-GPHN": "not imaged - no published localisation experiment",
"MLL-AF6": "IF + cell fractionation, nuclear"
}- If a cytoplasmic partner's wild-type pool is depleted by the fusion, a drug that stabilises or restores the fusion's partner-binding could worsen the cytoplasmic loss rather than relieve it.
- Gephyrin is required for molybdenum-cofactor synthesis and inhibitory synapse clustering, so any strategy directed at the gephyrin module of the fusion risks a neurological on-target toxicity with no haematological warning.
- Because MLL-GPHN transforms through a 15-residue oligomerisation module, there is almost no surface to drug and a competitive small molecule is implausible.
- Where does MLL-GPHN sit in a cell? The experiment has never been done for any gephyrin fusion.
- Do septin or gephyrin fusions deplete the cytoplasmic wild-type partner pool, as MLL-AF6 does for AF6?
- Does the de facto truncated MLL product of the alternative t(11;14) splice form exist as protein in a patient, and can it be distinguished from the chimera on a blot?
- E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
Immunofluorescent confocal microscopy of 293T cells transiently expressing MLL-SEPT6 or SEPT6 showed MLL-SEPT6 distributed as small speckles or dots in the nucleus, while overexpression of SEPT6 alone did not shift its physiological cytoplasmic localisation.
Dimerization of MLL fusion proteins and FLT3 activation synergize to induce multiple-lineage leukemogenesis · Ono R, et al. · The Journal of clinical investigation 2005What this evidence cannot show
Transient transfection of 293T, an overexpression system in a non-haematopoietic cell. The partner-alone control is in the same figure, which is the paper's strength; there is no comparison with a t(X;11) patient cell and no endogenous readout.
Evidence that argues against this (1)
- At physiological expression, low-complexity-domain interactions that control transcription form dynamic hubs without detectable phase separation, so a punctate appearance in a transiently transfected cell is weak evidence about the endogenous fusion. Chong S et al. 2018
- E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
GPHN was identified as a novel partner gene fused to MLL in a leukaemia with t(11;14)(q23;q24).
GPHN, a novel partner gene fused to MLL in a leukemia with t(11;14)(q23;q24) · Eguchi M, et al. · Genes, chromosomes & cancer 2001What this evidence cannot show
A single-case molecular cloning report; no protein localisation data. Abstract-level.
Evidence that argues against this (1)
- The independent report of the same translocation describes an alternative splice form in which the reading frame terminates shortly after the fusion point, producing a de facto truncated MLL with only seven extra residues - so not every t(11;14) transcript encodes a gephyrin-containing chimera. Kuwada N et al. 2001
- E3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
In one patient with de novo acute monoblastic leukaemia and t(11;14)(q23;q24), alternative splicing of the human gephyrin gene created two fusion cDNA forms - one in-frame to MLL exon 9 carrying the gephyrin C-terminus including the tubulin binding site, and one in which the reading frame terminated shortly after the fusion point, attaching only seven amino acids to the MLL N-terminus and producing a de facto truncated MLL.
t(11;14)(q23;q24) generates an MLL-human gephyrin fusion gene along with a de facto truncated MLL in acute monoblastic leukemia · Kuwada N, et al. · Cancer research 2001What this evidence cannot show
One patient; transcript-level evidence with no protein detection and no localisation. The functional significance of the truncated product is stated by the authors to be unclear. Abstract-level; no PMCID and no DOI in the PubMed record.
Evidence that argues against this (1)
- A truncated N-terminal MLL product was independently shown to be nuclear by immunofluorescence and cell fractionation, so the truncated species would be indistinguishable from the chimera by any localisation readout in use. Joh T et al. 1996
- E4strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
A 15-amino-acid tubulin-binding domain of gephyrin is necessary and sufficient for MLL-GPHN to transform haematopoietic progenitors in vitro and in vivo, confers oligomerisation capacity on MLL protein, and transduction of MLL-GPHN caused myeloid and lymphoid lineage leukaemias in mice - despite gephyrin lacking transcriptional activity.
The small oligomerization domain of gephyrin converts MLL to an oncogene · Eguchi M, et al. · Blood 2004What this evidence cannot show
A retroviral transduction model in mouse progenitors, which is an overexpression system by construction. The paper establishes oligomerisation, not localisation; no subcellular imaging or fractionation of MLL-GPHN is reported.
Evidence that argues against this (1)
- Forced oligomerisation as the conserved mechanism would predict no partner-specific compartment at all, which makes the absent MLL-GPHN localisation experiment a test of that model rather than a formality. Ono R et al. 2005
- E5moderate confidence 3 of 5: One strong study, or several smaller consistent ones.database
As of 2026-09-20 PubMed returns 3 records for `(MLL OR KMT2A) AND (GPHN OR gephyrin)`, with a query translation containing no injected [UID] term, and none of the three reports imaging or fractionation of the MLL-GPHN fusion protein.
PubMed E-utilities esearch, db=pubmed: (MLL OR KMT2A) AND (GPHN OR gephyrin), run 2026-09-20 · Not applicable - database query · NCBI E-utilities 2026What this evidence cannot show
Bounded by the symbol set. Gephyrin has older aliases (GPH, KIAA1385) not included here, and the retired-symbol lesson from the septin enumeration - where current symbols undercounted roughly tenfold - applies in principle. All three returned records were title-checked and accounted for individually.
Evidence that argues against this (1)
- A search built from the partner symbol alone can miss papers that frame the work at the level of the complex or the mechanism, which is how the NSL and Mediator absences in this project were produced; a mechanism-level query on oligomerisation might reach further. Eguchi M et al. 2004
- E6strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
MLL-AF6-rearranged cells relocate endogenous AF6 from the cytoplasm to the nucleus with consequent RAS activation, and silencing the fusion restores cytoplasmic AF6 and significantly reduces RAS-GTP levels.
MLL-AF6 fusion oncogene sequesters AF6 into the nucleus to trigger RAS activation in myeloid leukemia · Manara E, et al. · Blood 2014What this evidence cannot show
A different partner (AFDN) and a rearranged-cell system with a knockdown rescue, which is the strongest design in this class. Whether septin or gephyrin fusions do the same is untested.
Evidence that argues against this (1)
- Sept6-deficient mice show no gross abnormality and Sept6 loss does not alter MLL-SEPT6-induced disease, so for at least one septin the wild-type cytoplasmic pool appears dispensable and sequestration would carry no phenotype. Ono R et al. 2005
EB1 is a cytoplasmic partner like the septins and GPHN, yet the MLL-EB1 fusion protein is nuclear by IF (Fu 2005) - the same compartment question that entry frames
SEPTIN2 subnuclear localisation is NOT APPLICABLE - a cytoplasmic GTPase with no interphase nuclear pool in 143 SEPT2 or 71 SEPTIN2 records; UniProt's 'Chromosome, centromere, kinetochore' row is mitotic AND over-specifies its source, which says 'in close apposition to' (Q91 2026-09-22)
extends it with the GPHN case, which has no localisation experiment at all
no published image or fractionation of MLL-GPHN exists
MLL-SEPT6 is nuclear speckles while SEPT6 alone stays cytoplasmic, in the same figure
a 15-residue tubulin-binding oligomerisation module is what the fusion retains
the nuclear outcome is a property of retained KMT2A sequence, not of the partner
constrains the cytoplasmic-partner narrative by showing two partners are already nuclear in blood-lineage cells
MLL-GPHN is the untested case that could falsify it