AFF4-SDCBP (syntenin): the only prey-only row in this census typed 'direct interaction' - a proteomic phage-display hit on AFF4's C-terminal peptide, in vitro
also called AFF4-syntenin · AFF4-SDCBP · AFF4 C-terminal PDZ ligand · ProP-PD
A screen that displays the last few amino acids of every human protein on bacteriophage found AFF4's tail binding syntenin's peptide-binding module. It is a real biochemical measurement of a short motif, done entirely in a test tube with no cell.
The same thing, for a scientist
**Verdict: DECLINED as an interaction, RECORDED as a motif observation - and it is the only NON-HISTONE row among the 135 uncovered prey-only partners whose IntAct interaction type is `direct interaction`**. **CORRECTED before shipping:** this entry first claimed it was the only one outright. It is not - `H3C1` carries **ten** direct-interaction records (ITC x6, NMR, x-ray diffraction, pull down x2) from PMID 20541251, and H3C1 is the H3-tail substrate, not a partner [E1]. Method: **phage display**, in vitro, AFF4 as prey, intact-miscore 0.44, from **PMID 26787460** - Garrido-Urbani et al., *FEBS Lett* 2016, which profiles the **syntenin PDZ1-2 supramodule by proteomic peptide phage display (ProP-PD)** against a library displaying **C-terminal regions of the human proteome**, reports a preference for hydrophobic motifs and a tendency to recognise cryptic internal ligands, and validates one interaction (nectin-1) orthogonally [E2]. **So what was measured is that a peptide corresponding to AFF4's C-terminus binds syntenin PDZ1-2 on phage.** That is a genuine biophysical observation about a *motif*; it is not an observation about full-length AFF4 in a nucleus, and AFF4 was not among the interactions the authors validated.
Why `direct interaction` here means less, not more, than `association` elsewhere
PSI-MI's direct interaction (MI:0407) is the strongest interaction type in the vocabulary, and rule 93 records that the menin-KMT2A contact has seven such records through the REST endpoint. It is therefore tempting to read this row as the best-evidenced thing in the tail. It is not.
direct interaction describes what the assay demonstrates - two purified species in contact with no intermediary - and phage display satisfies that trivially, because one side is a synthetic peptide and the other a recombinant domain. It says nothing about whether the contact occurs in a cell. A ProP-PD library displays the terminal ~16 residues of every human ORF; the exercise is designed to enumerate candidate motifs, and its own authors describe it as complementary and of potential biological relevance.
The generalisable warning: in a tail ranked by evidence quality, interaction-type fields will float peptide-array and phage-display rows to the top. They are precise measurements of a small thing. A ranking that uses direct interaction as a quality signal without checking the method will systematically promote in vitro motif hits over in-cell complex purifications.
And the comparison that makes this concrete was already measured and written down. data/symbol-collisions.json records that the 11 KMT2A-H3C1 rows are ITC, NMR and X-ray of reader modules, and that all 11 concern the PHD3-bromodomain cassette, which EVERY KMT2A fusion DELETES 3. So the best-evidenced direct-interaction set in the whole uncovered tail - by method, by score and by record count - describes a contact the index fusion cannot make. The registry's conclusion is the one to carry: the reason to exclude substrates is the object, not the arithmetic - excluding the histone and ribosomal class moves coverage only from 43.8% to 46.1%, a 2.3-point change, so a denominator argument for excluding them fails and an object-level one succeeds.
The motif itself is the part worth keeping, and it is checkable
If AFF4's extreme C-terminus is a class-I or class-II PDZ ligand, that is a structural fact about AFF4 that does not depend on syntenin being the physiological partner. AFF4 is the scaffold of the super elongation complex; the C-terminal ALF homology / AHD region is the part that engages ELL and is the region KMT2A::AFF1's partner-side biology turns on.
Two things follow, neither of which needs new wet work to start:
- Which residues? The IntAct row carries no feature range, and the ProP-PD supplementary data were not opened here. The peptide sequence is recoverable from the paper and would say whether the motif is in the AHD or after it.
- Is it present in the fusion? KMT2A::AFF1 contains AFF1's C-terminus, not AFF4's. AFF1 and AFF4 are paralogues with a shared C-terminal architecture, so whether the same terminal motif exists in AFF1 - and therefore in the oncoprotein - is a sequence question answerable in minutes and not answered here.
That second question is the one this entry exists to raise. A PDZ-binding motif at the C-terminus of the retained partner moiety would be a protein-interaction handle on the fusion itself, which this project has almost none of.
| partner a | gene-aff4 |
|---|---|
| partner b | SDCBP |
| partner b uniprot | O00560 |
| partner b length AA | 298 |
| partner a uniprot | Q9UHB7 |
| partner a length AA | 1163 |
| verdict | DECLINED AS AN INTERACTION, RECORDED AS A MOTIF OBSERVATION |
| adjudicated by | q160, cycle 146, 2026-09-24 |
| motif b | AFF4 C-terminal peptide binding the syntenin PDZ1-2 supramodule; exact residues not recovered from the source |
| druggability | A PDZ-binding motif is a defined, short, peptide-mimetic-tractable element. Whether AFF1 carries the equivalent motif - and therefore whether KMT2A::AFF1 does - is unresolved and is the reason this row is kept. |
| n methods | 1 |
| n platforms | 1 |
| n publications | 1 |
| endogenous side | no - synthetic peptide against recombinant domain |
| intact miscore | 0.44 |
| interaction type mi | MI:0407 direct interaction |
| cell state | not applicable |
| epigenetics | None. |
| absence class | never-measured |
| absence class note | No cell-based experiment has tested AFF4 against syntenin, and no one has checked whether AFF1's C-terminus carries the same PDZ-ligand motif. The second is a sequence question and costs nothing. |
interface residues a
[]
structures
[]
ligands
[]
records
[ "AFF4-SDCBP, phage display, DIRECT INTERACTION, in vitro, AFF4 prey, miscore 0.44, PMID 26787460" ]
cell systems
[ "none - in vitro" ]
localization
{
"compartment": "not measured",
"subnuclear": [],
"shown_for_fusion": "NO - and the relevant unasked question is whether AFF1, not AFF4, carries the same C-terminal motif, since it is AFF1's C-terminus that the fusion retains."
}measurements
[
{
"quantity": "count",
"subject": "uncovered prey-only partners carrying any MI:0407 direct-interaction record",
"value": 2,
"units": "partners",
"n": 135,
"universe": "the 135 human prey-only partners with no entry, IntAct REST 2026-09-24",
"null": null,
"note": "H3C1 (10 records, ITC/NMR/x-ray, PMID 20541251) and SDCBP (1 record, phage display). H3C1 is the H3-tail SUBSTRATE, so among genuine partners the strongest interaction-type label in the tail belongs to an in-vitro phage-display hit."
}
]- A short linear motif is the easiest thing in a protein to target and the easiest to be wrong about: C-terminal peptides are over-represented in every display library artefact, and a motif that binds a promiscuous PDZ supramodule in vitro may bind nothing at cellular concentrations.
- Does AFF1 carry the same C-terminal PDZ-binding motif as AFF4? If it does, KMT2A::AFF1 ends in a peptide-mimetic-tractable element and this project has essentially no other protein-interaction handle on the fusion itself.
- How many rows across this whole graph are typed MI:0407 on the strength of a peptide-array or phage-display assay? Rule 93 uses direct-interaction counts as a quality signal for the menin-KMT2A contact; that usage is sound there and would be unsound if applied to the tail.
- Which AFF4 residues were displayed? Recoverable from the ProP-PD supplementary data, not retrieved here.
- Does the AFF4 C-terminal peptide motif lie inside a region the fusion retains? The H3C1 case in the collisions registry is the cautionary precedent: the best-evidenced structural contact in the tail maps to the PHD3-bromodomain cassette that every fusion deletes, so 'well measured' and 'relevant to the oncoprotein' came apart there and could here.
- E1established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database
IntAct REST returns one AFF4-SDCBP (Q9UHB7-O00560) record: detection method 'phage display', interaction type 'direct interaction', host 'In vitro', AFF4 as prey, intact-miscore 0.44, from PMID 26787460. Among the 135 uncovered human prey-only partners in this census, 11 records carry interaction type 'direct interaction': ten belong to H3C1 (P68431, from PMID 20541251 - ITC, NMR, x-ray diffraction and pull down) and one is this SDCBP row. SDCBP is therefore the only non-histone direct-interaction partner in the uncovered tail.
IntAct molecular interaction database, REST findInteractions on Q9UHB7, retrieved 2026-09-24 by q160 (partner SDCBP) · IntAct (EMBL-EBI) 2026What this evidence cannot show
One record. The row carries no feature range, so which AFF4 residues were displayed is not recoverable from IntAct and was not looked up in the paper's supplementary data. 'In vitro' as host organism is the tell that no cell was involved.
Evidence that argues against this (1)
- Being the only direct-interaction row could equally be read as this being the best-supported pair in the tail. That reading is defensible on the vocabulary and wrong on the method, which is exactly why the entry argues the point rather than asserting it. IntAct (EMBL-EBI) 2026
- E2established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper
PMID 26787460 (Garrido-Urbani et al., FEBS Lett 2016) profiles the syntenin PDZ1-2 supramodule by proteomic peptide phage display using a library displaying C-terminal regions of the human proteome; reports that the protein recognises a broad range of peptides with a preference for hydrophobic motifs and a tendency to recognise cryptic internal ligands; and validates the interaction with nectin-1 through orthogonal assays. AFF4 is not named in the abstract and is not the validated interaction.
Proteomic peptide phage display uncovers novel interactions of the PDZ1-2 supramodule of syntenin. · Garrido-Urbani S, Garg P, Ghossoub R, Arnold R, Lembo F, Sundell GN, Kim PM, Lopez M, Zimmermann P, Sidhu SS, Ivarsson Y · FEBS Lett 2016What this evidence cannot show
Abstract read in full; the ProP-PD hit tables and the AFF4 peptide sequence were NOT retrieved, so this entry cannot state which AFF4 residues bound. The authors' own statement that the domain 'recognises a broad range of peptides' is a promiscuity warning and is recorded as such.
Evidence that argues against this (1)
- ProP-PD is a well-controlled selection with sequence-level output, and a hit is a real binding event. Treating every unvalidated hit as noise would discard the method's entire purpose, which is to enumerate candidate motifs for later testing - which is what this entry recommends rather than dismisses. Garrido-Urbani S et al. 2016
- E3established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.dataset
data/symbol-collisions.json, added 2026-09-22 by king-cycle-57 from Q104, carries the row 'IntAct curates enzyme-substrate assays as interactions, so a partner census counts substrates as partners'. It records that H3C13 enters the partner union as a DOT1L partner by METHYLTRANSFERASE ASSAY - the H3K79 enzyme on its own substrate; that the 11 KMT2A-H3C1 rows are ITC, NMR and X-ray of reader modules and that all 11 concern the PHD3-bromodomain cassette, which every KMT2A fusion deletes; and that excluding the histone and ribosomal class (11 of 122 missing partners) moves coverage 43.8% to 46.1%, a 2.3-point change, which refutes a denominator argument for the exclusion. Its prescription is to filter the partner census by DETECTION METHOD, excluding enzymatic-assay and structural-fragment evidence, rather than by symbol class.
What this evidence cannot show
Project-internal registry, not a primary source. The PHD3-bromodomain claim about the 11 H3C1 rows was NOT independently re-checked against the IntAct feature ranges here, and this beat's own H3C1 tallies (11 direct-interaction records, 17 records total) were computed independently and happen to agree on the count of 11.
Evidence that argues against this (1)
- A contact the fusion cannot make is not worthless: wild-type KMT2A is still present in every KMT2A-rearranged cell and the PHD3-bromodomain cassette is intact on that allele. 'Irrelevant to the index fusion' and 'irrelevant to the disease' are different statements and the registry makes only the first. this project 2026
AFF4 C-terminal peptide binds syntenin PDZ1-2 on phage, in vitro, MI:0407
the open question: does AFF1, whose C-terminus the fusion retains, carry the same PDZ-ligand motif?
one of the 135 uncovered human prey-only partners