Where the break falls, and the three incompatible numbering frames that make it easy to get wrong
also called CALM-AF10 breakpoints · PICALM::MLLT10 fusion transcripts
The chromosome break can fall in several places, producing at least eight different versions of the fusion, and the versions behave differently. Worse, different papers count the amino acids in different ways, so the same position has three different numbers in the literature.
The same thing, for a scientist
**At least eight alternative PICALM::MLLT10 fusion transcripts are reported** [E1]. The canonical construct used throughout the murine literature joins **CALM 1-648** (all but the last four of 652) to AF10, and its numbering frame is stated as NCBI NP_009097.2, which equals UniProt Q13492-1 [E2]. Two breakpoint variants have been compared directly: **CALM-2091-AF10**, where AF10 replaces the last four residues of CALM, and **CALM-1926-AF10**, missing the C-terminal 165 residues of CALM; they differ in subcellular distribution and **alter the phenotype of the myeloid neoplasms they cause in mice** [E1]. Separately, patient transcripts split into **5' and 3' fusion types**, the 5' type retaining the AF10 extended LAP/PHD (PZP) region required for homo-oligomerisation, and the two types associate with different T-ALL maturation stages and different outcomes [E3]. **The numbering trap is the practical content of this entry** and is set out below.
Three frames, and never mix them
| entity | frame | length | used by |
|---|---|---|---|
| CALM/PICALM | UniProt Q13492-1 = NCBI NP_009097.2 | 652 | Conway et al., who state it explicitly; this graph |
| CALM/PICALM | short ubiquitous isoform, lacking 49 aa | ~603 | cell-biology literature |
| AF10/MLLT10 | UniProt P55197-1 | 1,068 | protein-mllt10-domains, the PDB co-crystals 6CKN/6CKO/6JN2 |
| AF10/MLLT10 | the 2026 review's frame ('AF10 contains 1027 amino acid residues'; fusion joins 'residues 81-1027') | 1,027 | the current review literature |
| AF10/MLLT10 | NCBI NP_001182555, where 'exon 11 = aa 234' | third frame | Conway et al. |
The same AF10 module is called OM-LZ 704-779 in P55197-1 numbering 4 and OM/LZ 677-758 in the review's frame - a difference of 27 residues at the N-terminal boundary and 21 at the C-terminal one, i.e. not a constant offset, so it cannot be repaired by adding a number. Any AF10 residue read out of a CALM-AF10 paper must have its frame established before it is used, and this graph asserts AF10 coordinates only in P55197-1.
This is the same family as the ASH2L 94-residue offset already logged in the queen brief: a number that is right in its own frame and wrong in yours.
Why the variants are not interchangeable
Two different axes of variation, and both have phenotypes:
- How much of CALM is retained. CALM-2091-AF10 and CALM-1926-AF10 differ by 165 C-terminal CALM residues - a span that contains the NES (544-553) and the reported dimerisation region (593-648). They localise differently and cause different myeloid phenotypes in mice 1.
- How much of AF10 is retained. 5' fusion transcripts keep the AF10 extended LAP/PHD (PZP) region; 3' transcripts do not. In the founding T-ALL series, 6 of 7 TCR-gamma-delta cases had 5' transcripts and 9 of 12 immature cases had 3' transcripts, and the two groups had very different outcomes 3.
If the second association is real, breakpoint position predicts maturation stage and outcome, and the mechanistic link is the PZP module that abolishes transformation when restored in cis. That connection has not been made in the literature and is the most testable loose end this beat found.
| gene5 | PICALM |
|---|---|
| gene3 | MLLT10 |
| cytoband | t(10;11)(p13;q14-21) |
| n reported transcript variants | 8 |
| canonical junction | CALM residues 1-648 of 652 joined to AF10; AF10 exon 11 in NP_001182555 numbering |
| frame mismatch example | OM-LZ is 704-779 in P55197-1 and 677-758 in the review's frame - 27 residues apart at one end and 21 at the other, so the discrepancy is NOT a constant offset and cannot be corrected arithmetically. |
| mechanism class | reciprocal translocation; breakpoint position varies and has phenotypic consequences |
variants compared directly
[
{
"name": "CALM-2091-AF10",
"description": "AF10 replaces the last four residues of CALM",
"localisation": "discrete cytoplasmic clusters; punctate nuclear under leptomycin B"
},
{
"name": "CALM-1926-AF10",
"description": "missing the 165 C-terminal residues of CALM, i.e. losing the NES and the reported dimerisation region",
"localisation": "diffuse cytoplasmic; diffuse nuclear under leptomycin B"
}
]patient transcript types
{
"5_prime": "retains the AF10 extended LAP/PHD (PZP) region; associated with TCR-gamma-delta phenotype and favourable outcome in the founding series",
"3_prime": "does not; associated with immature phenotype and poor outcome"
}numbering frames
{
"CALM_this_graph": "UniProt Q13492-1 / NP_009097.2, 652 aa",
"CALM_short_isoform": "approximately 603 aa",
"AF10_this_graph": "UniProt P55197-1, 1068 aa",
"AF10_review_frame": "1027 aa; the 2026 review states the fusion joins CALM to AF10 residues 81-1027",
"AF10_conway_frame": "NCBI NP_001182555; 'exon 11 = aa 234'",
"cDNA_nucleotide_labels": "the variant names 2091 and 1926 are CALM cDNA NUCLEOTIDE positions, not residues"
}- A junction-spanning MRD assay or a junction peptide designed from the canonical construct will miss the other seven transcript variants; ascertainment for this fusion is already halved by cytogenetic crypticity and a narrow molecular assay would halve it again.
- An AF10 residue number carried between a CALM-AF10 paper and a structure without converting the frame is wrong by 21-27 residues and will read as plausible.
- Does transcript type (5' versus 3') predict maturation stage and outcome in a modern RNA-sequenced cohort? A 2003 observation on 19 patients that would connect breakpoint position to the PZP brake if it held.
- What are the genomic breakpoint coordinates in GRCh38? No coordinate-level breakpoint data was located in this beat's reading; every description is at exon or cDNA level.
- Which of the eight reported transcripts occur in patients and at what frequency? The two directly compared in mice were chosen as representatives, not by prevalence.
- E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
Two CALM-AF10 breakpoint variants - CALM-2091-AF10, in which AF10 replaces the last four residues of CALM, and CALM-1926-AF10, missing the 165 C-terminal residues of CALM - differ in localisation (diffuse cytoplasmic versus discrete cytoplasmic clusters; diffuse versus punctate nuclear under leptomycin B) and alter the phenotype of the myeloid neoplasms they cause in mice; they represent two of at least eight alternative PICALM::MLLT10 fusion transcripts identified.
The clathrin-binding domain of CALM-AF10 alters the phenotype of myeloid neoplasms in mice. · Stoddart A, et al. · Oncogene 2012What this evidence cannot show
GFP-tagged constructs in 293T cells for the imaging, murine models for the phenotype. The '2091' and '1926' labels are NUCLEOTIDE positions in the CALM cDNA, not residue numbers - a fourth numbering convention in the same literature.
Evidence that argues against this (1)
- The variant labels 2091 and 1926 are CALM cDNA nucleotide positions rather than residue positions, which adds a fourth numbering convention to a literature that already carries three. A reader who takes them as residue numbers will place both breakpoints outside the 652-residue protein entirely. Stoddart A et al. 2012
- E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
The canonical CALM-AF10 chimera fuses all but the last four residues of CALM, with numbering referenced to the longest CALM isoform NCBI NP_009097.2, to AF10 exon 11 (residue 234 of NCBI NP_001182555).
A CALM-derived nuclear export signal is essential for CALM-AF10-mediated leukemogenesis. · Conway AE, et al. · Blood 2013What this evidence cannot show
This is the construct used in one laboratory's murine work, not a survey of patient breakpoints. Its value here is that it is the only source located that STATES its numbering frames for both partners, which is what makes the three-frame problem visible.
Evidence that argues against this (1)
- PUBLISHED CONFLICTING RESULT about using this construct as the reference frame. It is one laboratory's chimera, and how much of CALM is included is not cosmetic: in a mouse model built to vary exactly that, EXCLUDING the C-terminal portion of CALM required for optimal clathrin binding produced a myeloproliferative disease, while INCLUDING it produced acute myeloid leukaemia with changes in Pim1 and Crebbp. The 5' endpoint therefore selects the disease. A breakpoint entry that anchors its numbering to 'all but the last four residues of CALM' is anchoring to a construct whose phenotype is known to differ from constructs a few hundred residues shorter, and patient breakpoints are not all at that position. Stoddart A et al. 2012
- E3moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper
Among 12 immature CALM-AF10 T-ALLs, 9 showed 3' fusion transcripts, whereas 6 of 7 TCR-gamma-delta cases showed 5' fusion transcripts which retain the AF10 extended LAP/PHD domain necessary for homo-oligomerisation; the two groups had markedly different outcomes.
CALM-AF10 is a common fusion transcript in T-ALL and is specific to the TCRgammadelta lineage. · Asnafi V, et al. · Blood 2003What this evidence cannot show
19 patients across the two groups and a single cohort. The transcript-type assignment is by RT-PCR primer position rather than by sequencing the genomic breakpoint. Nobody appears to have retested the association with RNA sequencing.
Evidence that argues against this (1)
- The clinical half of the transcript-type association is contested: a paediatric T-ALL series found no significant outcome difference between PICALM-MLLT10-positive and -negative children (5-year EFS 71.4% versus 62.5%), so any prognostic weight assigned to breakpoint position rests on a single 19-patient stratification within one cohort. Lo Nigro L et al. 2013
- E4established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database
The DOT1L co-crystal constructs place the MLLT10 OM-LZ at residues 704-779 (6CKN, 6CKO) and 699-782 (6JN2) in UniProt P55197-1 numbering.
What this evidence cannot show
Per the brief's PDB rules, residue indices read straight off a deposit are often offset from the UniProt frame; these are recorded as the numbering this graph's `protein-mllt10-domains` entry already uses, and the conflict with the review's 677-758 call is the point of this entry rather than an incidental note.
Evidence that argues against this (1)
- METHODOLOGICAL DEFEATER: a deposit cannot set a module's functional boundary. A crystallised OM-LZ construct is chosen for diffraction and solubility, so 704-779 and 699-782 are the spans that gave crystals, not the spans required for function. The functional limits of this module were set the other way, by deletion in transformation assays - the AF10 OM-LZ fragment that suffices to induce AML in mice when fused to a CALM C-terminus is defined by a colony and transplantation readout, and there is no reason for that boundary to coincide with a crystallisable one. The three-frame problem this entry documents is real; adding a fourth frame from crystallography and treating it as authoritative would make it worse. Deshpande AJ et al. 2011
where the break falls, and the numbering frames
- related to →PICALM / CALM - domain architecture, residue coordinates, and which parts the fusion keeps
CALM coordinates asserted in Q13492-1
AF10 coordinates asserted in P55197-1, which is not the frame the CALM-AF10 literature uses
5' versus 3' transcript type associates with maturation stage and outcome