ABL1 — domain architecture and residue coordinates (with the 1a/1b numbering trap)
also called ABL1 · c-Abl · ABL · JTK7 · P00519
ABL1 is an enzyme that adds phosphate tags to other proteins, and it normally keeps itself switched off by folding a greasy tail back against its own working face. The leukaemia fusion cuts that tail off, which is why the enzyme never switches off again, and because two rival ways of counting ABL1's residues differ by exactly nineteen, every number here has to say which one it uses.
The same thing, for a scientist
ABL1 (UniProt P00519, canonical isoform 1a, 1,130 aa) comprises an N-terminal cap (1-60), SH3 (61-121), SH2 (127-217), a tyrosine kinase domain (242-493) with an activation loop at 381-405, three nuclear localisation signals, a DNA-binding region (869-968), an F-actin-binding domain (953-1130) and a nuclear export signal (1090-1100). Autoinhibition requires the myristoylated N-terminus of isoform 1b docking into a pocket on the kinase C-lobe. Two numbering systems exist and differ by exactly 19: we verified by direct alignment that 1a residue 27 equals 1b residue 46 and that the isoforms share an identical 1,104-residue C-terminal sequence. We then read the actual amino acid at every standard clinical mutation position in both sequences and found that M244, L248, G250, Y253, E255, T315, F317, M318, M351, F359, H396 and F486 all match **isoform 1a**, i.e. the canonical P00519 sequence. The widely repeated claim that the T315I convention is isoform-1b numbering is wrong: position 315 of isoform 1b is a proline, and the gatekeeper in 1b numbering is T334. All coordinates in `data.domains` are 1a/P00519 numbering, which is also the clinical convention, so standard mutation annotations need no conversion.
This is a prediction, not a measurement. No one has seen this protein in this shape. The colours say how much the software trusts itself, and that is worth reading before anything else — the orange stretches are places the model is effectively saying this part has no fixed shape.
I have never read one of these pictures before
A protein is one long chain of amino acids that folds up. The ribbon traces that chain from one end to the other; it is not a wire, it is a simplification of where the backbone runs. Drag to rotate, scroll to zoom.
Domains are the stretches of chain that do a particular job, and they are what the coloured blocks above the picture mark out. Click one and the view flies to it. In this project the important question about any domain is usually whether the fusion still has it — the labels say lost where it does not.
Anything in yellow sticks is a bound small molecule, which is usually the drug.
The 1a/1b offset, resolved exactly — and a correction1 deeper
This is the most dangerous number in the comparator set, so it was checked by direct sequence alignment and by reading the actual amino acid at each disputed position, not by recall 1.
- Isoform 1a = UniProt canonical P00519-1 = 1,130 aa. Its first 26 residues are 1a-specific.
- Isoform 1b = P00519-2 = 1,149 aa. Its first 45 residues are 1b-specific and carry the Gly2 myristoylation site.
- The sequences are identical from 1a residue 27 / 1b residue 46 onwards, a shared stretch of 1,104 residues. The offset across the shared region is exactly +19.
The correction. It is often asserted that the clinical T315I convention is isoform-1b numbering. It is not. Reading residues straight out of both sequences:
| Mutation name | Residue at that position in 1a (canonical P00519) | Same residue in 1b |
|---|---|---|
| M244 | M | 263 |
| Y253 | Y | 272 |
| E255 | E | 274 |
| T315 | T | 334 (position 315 of 1b is a proline) |
| F317 | F | 336 |
| M318 | M | 337 |
| M351 | M | 370 |
| F359 | F | 378 |
| H396 | H | 415 |
Every standard kinase-domain mutation name resolves correctly against isoform 1a, which is the UniProt canonical sequence. So the clinical convention and the reference sequence already agree and no conversion is needed for standard mutation names. The offset matters in the opposite direction: a minority of older structural papers number against 1b, where the gatekeeper is T334 and the kinase domain is 261-512.
A useful coincidence at the fusion junction
UniProt annotates the BCR::ABL translocation site between ABL1 residues 26 and 27 1 — precisely the point where the 1a and 1b sequences converge, because the a2 exon boundary sits at the end of the isoform-specific first exon. The ABL1 part of BCR::ABL1 is therefore the same peptide whichever isoform you start from, and the +19 offset applies as a clean constant rather than a piecewise mess. It also means the 1b-specific myristoylated N-terminus is not merely displaced by BCR; it lies outside the fusion entirely.
What autoinhibition needs, and what the fusion removes
Nagar and colleagues showed crystallographically that c-Abl's autoinhibited state is built by the N-terminal myristoyl group of isoform 1b binding a pocket in the kinase C-lobe, which induces the conformational change that lets SH3 and SH2 clamp onto the kinase domain — an assembly closely similar to inactive Src but with differences that explain why imatinib inhibits Abl and not Src 2.
In BCR::ABL1 the entire 1b-specific first exon, myristoyl site included, is replaced by BCR. The latch is not merely loosened; the peg is gone. SH3 and SH2 are both retained, but with nothing to clamp against them they become docking surfaces instead of brakes — and the retained BCR region 197-385 supplies an intramolecular SH2 ligand. That is the structural statement of why this fusion is constitutively active.
Where the drug binds and where resistance lands
Imatinib binds the kinase domain in its inactive DFG-out conformation (PDB 2HYY, entry title verified as 'Human Abl kinase domain in complex with imatinib'). The first clinical resistance mechanism found was a single substitution at the gatekeeper threonine that forms a critical hydrogen bond with the drug, plus, in a minority, amplification of the fusion gene itself 3. In 1a numbering that threonine is residue 296; almost nobody writes it that way.
The practical consequence for this graph: the viewer's residue ruler is 1a, so a resistance annotation imported from a clinical dataset will be 19 residues to the right of where it belongs unless converted.
| uniprot | P00519 |
|---|---|
| length AA | 1130 |
| gene symbol | ABL1 |
| isoform note | All coordinates below are UniProt canonical P00519-1 = isoform 1a = 1,130 aa. Verified by direct alignment: 1a residues 1-26 and 1b residues 1-45 are isoform-specific; the sequences are identical from 1a-27 / 1b-46 onwards (1,104 shared residues), so 1b = 1a + 19 across the shared region. CORRECTION to a widespread belief: the standard clinical mutation names (M244V, G250E, Y253H, E255K, T315I, F317L, M351T, F359V, H396R, F486S) are isoform-1a names and resolve correctly against this canonical sequence. Residue 315 of isoform 1b is a proline; the gatekeeper in 1b numbering is T334. No conversion is required for standard clinical annotations, only for the minority of older structural papers that number against 1b. |
| numbering offset 1b minus 1a | 19 |
| gatekeeper residue 1a | 315 |
| gatekeeper residue 1b | 334 |
| assumed fusion | BCR::ABL1 p210 (e14a2/b3a2). The ABL1 contribution is residues 27-1130 in 1a numbering (= 46-1149 in 1b numbering) and is identical in p190 and p230. |
| alphafold ID | AF-P00519-F1 |
| domain coordinate caveat | SH3, SH2 and the kinase domain are pinned by dozens of crystal structures and are reliable to ~+/-2 residues (UniProt SH3 61-121 vs PDB constructs 59-122; SH2 127-217 vs construct 112-232; kinase 242-493 vs constructs 227-515). The CAP and DNA-binding 'domains' are functional regions with no experimental structure and boundaries that should be treated as soft to tens of residues; the DNA-binding interval at 869-968 has no support outside UniProt. The F-actin-binding domain is NOT without structure and its stored start is the single largest coordinate disagreement in this node: InterPro and SMART give 1004-1130, CATH 1024-1130, Pfam 1026-1130 and the NMR structure 1ZZP 1007-1130, against UniProt's 953. Treat 953-1003 as functional-region padding, not as fold. Every number here is isoform-1a, which is also the clinical mutation-naming convention; add 19 only for a source that explicitly uses 1b numbering. |
| fusion boundary residue | 27 |
| fusion boundary note | ABL1 contributes residues 27-1130 (1a) = 46-1149 (1b). UniProt independently annotates the translocation site at 26/27, which is also exactly where the 1a and 1b sequences converge. |
verified mutation residues 1a
{
"M244": "M",
"L248": "L",
"G250": "G",
"Y253": "Y",
"E255": "E",
"T315": "T",
"F317": "F",
"M318": "M",
"M351": "M",
"F359": "F",
"H396": "H",
"F486": "F"
}PDB IDS
[ "1OPL", "2HYY", "2ABL", "1BBZ", "1ZZP" ]
domains — the numbers behind the chart
[
{
"name": "CAP (N-terminal cap)",
"start": 1,
"end": 60,
"note": "TRUNCATED, and this is the whole point: BCR replaces residues 1-26 (in isoform 1b, residues 1-45 including the Gly2 myristoylation site). Residues 27-60 survive but the myristoyl latch that locks the kinase off does not. Marked lost.",
"fate": "lost"
},
{
"name": "SH3",
"start": 61,
"end": 121,
"note": "retained; in autoinhibited c-Abl it clamps the kinase N-lobe, in the fusion it has nothing to clamp.",
"fate": "retained"
},
{
"name": "SH2",
"start": 127,
"end": 217,
"note": "retained; also bound intramolecularly by BCR 197-385 in the fusion.",
"fate": "retained"
},
{
"name": "PK (tyrosine kinase domain)",
"start": 242,
"end": 493,
"note": "the imatinib site. The gatekeeper threonine T315 is residue 315 of THIS (1a) sequence; in 1b numbering it is T334 and the domain is 261-512. CATH-Gene3D splits the fold into 229-317 and 319-542, i.e. 49 residues past UniProt's end at 493; the imatinib-bound construct 2HYY spans 228-500 [E8].",
"fate": "retained"
},
{
"name": "A-loop (kinase activation loop)",
"start": 381,
"end": 405,
"note": "= 400-424 in 1b numbering. H396 lies inside it and is residue 396 of this (1a) sequence.",
"fate": "retained"
},
{
"name": "NLS1",
"start": 605,
"end": 609,
"fate": "retained"
},
{
"name": "NLS2",
"start": 709,
"end": 715,
"fate": "retained"
},
{
"name": "NLS3",
"start": 762,
"end": 769,
"fate": "retained"
},
{
"name": "DNA-binding region",
"start": 869,
"end": 968,
"note": "functional assignment, no structure; lies inside a long predicted-disordered stretch. Single-source coordinate: no profile database annotates anything in ABL1 between residues 509 and 1004 [E7]. It also overlaps the stored FABD by sixteen residues.",
"fate": "retained"
},
{
"name": "FABD (F-actin-binding domain)",
"start": 953,
"end": 1130,
"note": "PDB 1ZZP covers 1007-1130. BOUNDARY DISPUTE: UniProt's functional region is 953-1130, but InterPro IPR015015 and SMART give 1004-1130, CATH-Gene3D 1024-1130, Pfam PF08919 1026-1130 and the solution structure 1ZZP covers 1007-1130 [E6]. The folded domain is roughly the last 124 residues; 953-1003 is functional-region padding and overlaps this node's DNA-binding region at 869-968.",
"fate": "retained"
},
{
"name": "NES (nuclear export signal)",
"start": 1090,
"end": 1100,
"note": "inside the FABD; its retention is why BCR::ABL1 is cytoplasmic despite having three NLSs.",
"fate": "retained"
}
]counter evidence pass
{
"queen": "Q2b-atlas2",
"cycle": 31,
"date": "2026-09-19",
"checked": [
"UniProt P00519 live record (primary gene name ABL1 confirmed, 1130 aa; all eleven stored intervals match UniProt exactly, including the BCR-ABL translocation site at 26/27)",
"InterPro entry/all for P00519",
"SIFTS mappings for 1OPL, 2HYY, 2ABL, 1BBZ, 1ZZP"
]
}- A confidently wrong number here is worse than no number, and the trap runs the opposite way from how it is usually stated: applying a -19 correction to T315 because someone believed the clinical convention was 1b would land the gatekeeper on a proline and corrupt any structure-based argument built on the figure.
- Imatinib and its successors bind a pocket that wild-type ABL1 also has; ABL1 has real physiological roles in DNA damage response and actin dynamics, so on-target toxicity is not hypothetical.
- Resistance to ATP-site inhibitors was solved by an allosteric myristoyl-pocket drug, but that pocket exists only because the fusion lost its own myristoyl ligand — a reminder that the therapeutic opportunity here was created by the very truncation that causes the disease, and the same logic will not transfer to fusions that lose nothing druggable.
- Does anything in the retained C-terminal half of ABL1 (the DNA-binding and F-actin-binding regions, none of which has an experimental structure) contribute to transformation, or is it inert cargo? A fusion truncated after the kinase domain would answer this.
- The three retained NLSs are overridden by the single retained NES plus the BCR moiety. Is nuclear entry of BCR::ABL1 pharmacologically inducible, and would trapping the fusion in the nucleus kill the cell?
- How much of the published assertion that ABL1 clinical numbering is isoform-1b is simply repeated error? We could not reproduce it: every standard mutation name resolves against 1a. Auditing resistance-variant databases for silent 1a/1b mis-conversion is a concrete, cheap analysis.
- Where does the F-actin-binding domain actually start? UniProt says 953, the structure and every profile model say ~1004-1026. The answer decides whether the p210 BCR::ABL1 fusion carries a 178-residue actin-binding module or a 124-residue one plus 50 residues of linker, and the difference is visible in any domain figure the graph draws.
- E1established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database
ABL1 canonical P00519-1 (isoform 1a) is 1,130 aa with CAP 1-60, SH3 61-121, SH2 127-217, kinase 242-493, activation loop 381-405, NLS motifs at 605-609, 709-715 and 762-769, DNA-binding 869-968, F-actin-binding 953-1130 and NES 1090-1100; isoform 1b (P00519-2) is 1,149 aa and the BCR::ABL translocation site is annotated between residues 26 and 27.
UniProtKB entry P00519 (ABL1_HUMAN): reviewed sequence, length and positional feature annotations · UniProt Consortium · UniProtKB/Swiss-Prot 2026What this evidence cannot show
Isoform lengths were re-derived by downloading both isoform FASTA sequences and the +19 offset and the 1a-27 = 1b-46 convergence point were computed directly, not taken from prose. The 'DNA-binding' and 'F-actin-binding' regions are functional assignments with no structure covering them.
Evidence that argues against this (2)
- The F-actin-binding bar at 953-1130 is the weakest coordinate in this node and it is 51-73 residues wider at the N-terminus than every independent source: InterPro/SMART 1004-1130, CATH 1024-1130, Pfam 1026-1130, and the NMR structure 1ZZP 1007-1130. It also overlaps the DNA-binding region at 869-968. We keep UniProt's value because the node's stated convention is UniProt numbering and the region is functional rather than structural, but a figure drawn from it will show a 178-residue FABD where the folded domain is about 124. InterPro Consortium; EMBL-EBI 2026
- CATH-Gene3D splits the kinase domain into an N-lobe at 229-317 and a C-lobe at 319-542, so the structural classification runs 49 residues past UniProt's kinase end at 493. Nothing in this node depends on it, but a domain bar ending at 493 stops short of the C-lobe as CATH defines it. InterPro Consortium; EMBL-EBI 2026
- E2established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper
c-Abl autoinhibition requires the N-terminal myristoyl modification of isoform 1b binding the kinase domain, which induces the conformational change allowing SH3 and SH2 to dock; the assembly resembles inactive Src but differs in ways that explain imatinib's selectivity.
Structural basis for the autoinhibition of c-Abl tyrosine kinase · Nagar B, Hantschel O, Young MA, et al. · Cell 2003What this evidence cannot show
Crystal structures of a construct, not of full-length protein in cells. PDB 1OPL, deposited with this work, is mapped by UniProt to residues 6-512 and uses 1b numbering in the original paper.
Evidence that argues against this (2)
- The structure this rests on does not show the myristoylated N-terminus attached to anything. In PDB 1OPL the myristoyl group is deposited as a FREE myristic-acid molecule (nonpolymer entity, comp_id MYR) together with a pyridopyrimidinone inhibitor at 3.42 A, and PDBe polymer coverage shows the ABL1 chain has coordinates only from author residue 81 in chain A and 140-237 plus 252-518 in chain B - so the entire CAP region UniProt annotates at 1-60, the region the lipid is supposed to be attached to, is unmodelled in both copies. The companion higher-resolution entries 1OPJ (1.75 A) and 1OPK (1.8 A) contain only the isolated kinase domain. The mechanism is inferred from a lipid modelled in a pocket, not from an observed covalent tether. PDBe / SIFTS et al. 2026
- Imatinib's selectivity was explained three years earlier from a construct with no SH3, no SH2 and no myristoyl group at all. The isolated Abl catalytic domain bound to an STI-571 variant attributed both the high affinity and the specificity to the catalytic domain's own unphosphorylated activation-loop conformation, distinct both from active kinases and from the inactive form of the closely related Src kinases. The autoinhibited SH3-SH2-kinase assembly may be real without being what selectivity runs through. Schindler T et al. 2000
- E3established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper
Clinical imatinib resistance is caused in most cases by a single substitution of the kinase-domain threonine that hydrogen-bonds the drug, and in a minority by BCR-ABL gene amplification.
Clinical resistance to STI-571 cancer therapy caused by BCR-ABL gene mutation or amplification · Gorre ME, Mohammed M, Ellwood K, et al. · Science 2001What this evidence cannot show
Nine patients with advanced-phase disease. The paper names the residue T315; we verified that residue 315 of the canonical P00519 (isoform 1a) sequence is a threonine and that residue 315 of isoform 1b is a proline, so this is 1a numbering and needs no conversion for this node.
Evidence that argues against this (1)
- Once cohorts got past single digits, T315I stopped being the majority event and stopped being even the most common one. Screening 171 patients failing imatinib found 66 mutations at 23 different amino acids in 62 patients (36% of those screened), and the largest single class was P-loop mutations (n=24, 36% of the 66), not T315I. A separate 370-patient series found mutations in 127 of 297 evaluable patients (43%) with seven residues - M244V, G250E, Y253F/H, E255K/V, T315I, M351T, F359V - together accounting for 85% of resistance-associated mutations. The amplification sub-claim is also over-weighted by an n=9 founding series: genomic BCR-ABL amplification by FISH was found in only 2 of 32 evaluated among 66 imatinib-resistant patients. Jabbour E et al. 2006
- E4established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database
An AlphaFold model of full-length ABL1 exists as AF-P00519-F1 covering residues 1-1130, and experimental structures cover the SH3-SH2-kinase module (e.g. 1OPL, residues 6-512) and the isolated kinase domain with imatinib (2HYY).
AlphaFold Protein Structure Database entry for P00519 (model AF-P00519-F1) · Jumper J, Varadi M, et al.; EMBL-EBI/DeepMind · AlphaFold DB 2026What this evidence cannot show
The C-terminal half of ABL1 (UniProt annotates disorder across 518-996) is predicted with low confidence and has no experimental structure; the DNA-binding and F-actin-binding regions should be drawn as regions, not folds. The 1OPL and 2HYY entry titles were retrieved from RCSB and confirmed.
Evidence that argues against this (1)
- Coverage is not confidence: AlphaFold's own per-model metrics say half of that model should not be read as structure. The fetched record for AF-P00519-F1 (v6) gives a global pLDDT of 63.38 with 49.2% of residues in the very-low band (pLDDT below 50), 4.5% low, 9.4% confident and 36.9% very high - 557 of 1,130 residues in the band AlphaFold explicitly advises against interpreting. The two experimental structures do not compensate: 1OPL is 3.42 A with residues 6-61 unmodelled, and 2HYY's ordered range is author residues 235-274 plus 276-498 out of a 228-500 construct. AlphaFold DB; EMBL-EBI / Google DeepMind 2026
- E5strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.review
In BCR::ABL1 the ABL1 contribution begins at the a2 exon, i.e. canonical residue 27, in both p190 and p210.
The molecular biology of chronic myeloid leukemia · Deininger MW, Goldman JM, Melo JV · Blood 2000What this evidence cannot show
A 2000 review, cited by PMID (no DOI indexed). The residue-27 assignment is corroborated independently by the UniProt translocation-site annotation at 26/27.
Evidence that argues against this (1)
- A real minority of fusions skip ABL1 exon a2 entirely and splice BCR straight onto exon a3, so the ABL1 contribution starts later than residue 27. In a series of 52 CML patients carrying eight types of atypical BCR::ABL1 transcript, e13a3/e14a3 was one of the three commonest atypical classes with 11 patients, alongside e19a2 (n=17) and e1a2 (n=6), and the paper puts atypical transcripts overall at 0.3-4.6% of CML. The claim should be scoped to the canonical transcripts rather than asserted for BCR::ABL1 generally. Wang H et al. 2025
- E6strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.database
Every profile database and the only experimental structure of the region place the ABL1 F-actin-binding domain 50-73 residues later than the 953 this node stores: InterPro IPR015015 and SMART SM00808 give 1004-1130, CATH-Gene3D G3DSA:1.20.120.330 gives 1024-1130, Pfam PF08919 gives 1026-1130, and the solution structure PDB 1ZZP ('Solution structure of the F-actin binding domain of Bcr-Abl/c-Abl') maps to P00519 residues 1007-1130.
InterPro member-database domain matches for UniProt P00519 (Pfam, PROSITE, SMART, CDD, CATH-Gene3D, SUPERFAMILY) · InterPro Consortium; EMBL-EBI · InterPro 2026What this evidence cannot show
Database records read back 2026-09-19; no full text read for this item. UniProt's 953-1130 is a curated FUNCTIONAL region taken from actin-binding mapping literature, not a fold boundary, so the two are not strictly in conflict - but they are drawn as the same kind of bar in this graph's viewer, and 953-1130 also overlaps this node's own DNA-binding region at 869-968 by sixteen residues. The 1ZZP construct boundary is itself a construct choice and not a measurement of where the fold begins.
Evidence that argues against this (1)
- The structure does not support the earliest of those numbers, and the three database calls span 22 residues among themselves. SIFTS maps 1ZZP chain A to P00519 1007-1130, but PDBe polymer coverage shows coordinates only for residue numbers 22-130 across the 10 deposited NMR models, so order begins around UniProt 1024 and residues 1007-1023 of the construct have no coordinates at all. That makes CATH-Gene3D (1024) and Pfam (1026) the numbers the experiment actually supports and leaves InterPro IPR015015 / SMART SM00808's 1004 start 20 residues beyond any density. UniProt's own record undercuts its 953 start from the other direction: Region 518-996 is annotated Disordered, so UniProt calls residues 953-996 of its own F-actin-binding region disordered. PDBe / SIFTS et al. 2026
- E7strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.database
No profile database annotates any domain in ABL1 between residue 509 and residue 1004, so the DNA-binding region stored at 869-968 has no support outside UniProt's own curated functional annotation.
InterPro member-database domain matches for UniProt P00519 (Pfam, PROSITE, SMART, CDD, CATH-Gene3D, SUPERFAMILY) · InterPro Consortium; EMBL-EBI · InterPro 2026What this evidence cannot show
Absence of a profile match is weak evidence about a real but unstructured functional region, and DNA binding by the ABL1 C-terminus is reported in the primary literature; this item records only that the 869-968 interval is a single-source coordinate, not that the function is doubtful. Database record, retrieved 2026-09-19.
Evidence that argues against this (1)
- UniProt's own record is not a second, independent vote for 869-968 - it contradicts itself there. The fetched P00519 entry annotates Region 518-996 as Disordered, which completely contains the DNA-binding region 869-968, so the same curated record simultaneously asserts a functional DNA-binding module and predicted disorder across all 100 residues of it. The 40 InterPro matches confirm the profile gap (PANTHER PTHR24418 / IPR050198 stop at 509; nothing resumes until IPR015015 at 1004), so the interval has exactly one supporting annotation and that annotation is internally inconsistent with its neighbour in the same file. UniProt Consortium 2026
- E8established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database
Deposited construct boundaries bracket but do not confirm the catalytic-module annotations: SIFTS maps 1OPL to P00519 6-512, 2HYY (imatinib-bound kinase) to 228-500, 2ABL to 57-218 and 1BBZ to 64-121, against stored SH3 61-121, SH2 127-217 and kinase 242-493.
SIFTS UniProt residue-level mappings for PDB entries 1ZZP, 1OPL, 2HYY, 2ABL, 1BBZ (construct ranges read back from the PDBe mappings API) · PDBe / SIFTS; EMBL-EBI · PDBe SIFTS 2026What this evidence cannot show
Construct ranges from the SIFTS mapping, which reports the deposited sequence range rather than the observed model; database record read 2026-09-19, no full text. Construct choices are made for crystallisability and routinely overshoot a domain at both ends, so these bracket the annotation rather than testing it.
Evidence that argues against this (1)
- Construct ranges and modelled ranges are different measurements, and for the biggest construct they differ by 56 residues. PDBe polymer coverage gives 1OPL chain A coordinates only for author residues 81-531 and chain B for 140-237 plus 252-518, against a SIFTS construct of UniProt 6-512 - so the first 56 UniProt residues of the deposited construct (6-61) are absent from both copies, and chain B additionally loses 238-251. For 2HYY the ordered range is author 235-274 and 276-498 against a 228-500 construct. A construct range therefore does not even bracket the annotation reliably: it brackets what was cloned, and what was cloned was not all seen. PDBe / SIFTS et al. 2026
Counter-evidence added cycle 41: imatinib resistance is not mostly T315I. Screening 171 patients failing imatinib found 66 mutations at 23 residues, the largest class being P-loop mutations (n=24, 36%), not T315I (Jabbour 2006, PMID 16855631); and genomic BCR-ABL amplification was found in only 2 of 32 evaluated among 66 resistant patients. Also: the autoinhibition structure 1OPL has NO coordinates for ABL1 residues 6-61, the region the myristoyl is supposed to tether.
Boundary dispute (InterPro P29590-style check on P00519): the F-actin-binding domain this node draws from 953 starts at 1004-1026 in InterPro, Pfam, SMART and CATH, and the NMR structure 1ZZP covers only 1007-1130. The fusion's actin-binding module is ~124 residues of fold plus ~50 of functional-region padding, not a 178-residue domain.
BCR::ABL1, t(9;22)
3-prime partner domain architecture; the autoinhibitory N-terminal cap is lost and the kinase domain retained
loss of the myristoylated N-terminal cap removes the kinase own off-switch
residue coordinates for the ATP pocket and the Thr315 gatekeeper, in verified isoform-1a numbering
ABL1 a2 exon joins the BCR major or minor breakpoint cluster
the retained kinase domain (242-493) carries an evolved, druggable ATP pocket
- analogous to →RARA — domain architecture and residue coordinates
both fusions retain an intact, pre-existing small-molecule binding pocket; neither pocket was created by the fusion
3' partner domain coordinates
SFPQ exon 9 to ABL1 exon 4 in every reported case; the intron-4 break leaves the chimaera without the ABL1 SH2 domain