The binary ladder for AFF1: does the index fusion's partner touch P-TEFb without AFF4, and does it touch AFF4 at all?
also called AFF1 ITC against P-TEFb · AFF4-null AFF1 immunoprecipitation · turning sixteen spokes into one number
Buy or make the pieces, mix two at a time, and measure whether they stick. The whole field's picture of how the AF4 protein plugs into the gene-switching machine comes from experiments that pulled everything out at once, so nobody knows which piece it actually holds on to.
The same thing, for a scientist
Two arms, neither requiring anything the field does not already have. **Arm A, purified components:** recombinant AFF1 fragments spanning the region paralogous to AFF4's P-TEFb-binding segment, titrated against purified CDK9/cyclin T1 by ITC and SPR, with the equivalent AFF4 fragment as the positive control and the AFF4-P-TEFb structures as the calibration [E1]. **Arm B, cells:** AFF1 immunoprecipitation from parental and AFF4-knockout cells, blotting CDK9, cyclin T1, ELL2 and MLLT1, with AFF4 IP as the reciprocal and an AFF1-knockout lysate as the antibody control. Arm A gives a number or a bounded absence; arm B gives topology in the cell. Together they convert sixteen spoke-expanded IntAct records into either a dissociation constant or a bounded negative [E2], and they discriminate the two topologies in `hyp-aff1-reaches-ptefb-only-through-aff4`. Cost tier low-to-medium, timeline weeks, and the cheapest step of all costs nothing: a full-text sweep for an existing binary measurement should be done first, because it could refute the hypothesis before a pipette is lifted.
Step 0, free, and mandatory before anything else
Sweep the full texts, not the abstracts. Brief rule 80: a PubMed zero is bounded by abstracts, and the measurement a mechanism graph wants is almost always a supporting panel. A single ITC or GST-pulldown-from-purified-components panel in any SEC paper since 2010 would settle the question for free and would make both arms below unnecessary.
Search Europe PMC full text for AF4/AFF1 together with each of isothermal titration, surface plasmon, Kd, purified recombinant, GST pulldown, and do the same for the retired symbol AF4 and for MLLT2, which is AFF1's previous official symbol and reaches literature the current symbol does not. State the name list in the caveat.
If the sweep finds a measurement, this entry becomes a citation and the hypothesis dies cheaply. That is the best outcome available and it should be sought before the reagents are ordered.
Arm A - purified components, two at a time
Constructs. Align AFF1 (P51825, 1,210 aa) against AFF4 (Q9UHB7, 1,163 aa) and identify the segment paralogous to the AFF4 region resolved in 4IMY against CDK9/cyclin T1 1. Express that AFF1 segment, the equivalent AFF4 segment as a positive control, and a charge-matched scramble as a negative.
Analyte. Purified CDK9/cyclin T1 heterodimer, which is commercially available and is the same object used in the 4IMY work.
Readouts, in order of decisiveness:
| method | gives | why include it |
|---|---|---|
| ITC | Kd, stoichiometry, enthalpy | label-free, solution, no surface artefact; this is the number |
| SPR | Kd, kon/koff | catches weak, fast-off binding that ITC's concentration range can miss |
| analytical SEC | co-migration | a sanity check that both components are folded and monodisperse |
The controls that decide whether a negative means anything. The AFF4 fragment must bind in the same session, on the same protein preparation, at a Kd consistent with the published AFF4/P-TEFb interaction. Without that positive control an AFF1 negative is uninterpretable - this project's rule 67 in biophysical form: never trust a zero from an assay you have not shown can return a one.
Bounded negative. If AFF1 does not bind, report the concentration ceiling reached and state the Kd bound explicitly ("no binding detectable to 200 µM, so Kd > 200 µM"), not "no interaction".
Arm B - topology in a cell
Cells. A KMT2A::AFF1 line (SEM or RS4;11) and a non-rearranged control, each with and without AFF4 knockout. AFF4-null cells are viable, which is what makes this arm possible.
Design. Immunoprecipitate AFF1; blot CDK9, cyclin T1, ELL2, MLLT1 and AFF4. Reciprocally IP AFF4 and blot AFF1 and the same panel. Include an AFF1-knockout lysate as the antibody specificity control, which is the control most often missing in this literature.
| result | reading |
|---|---|
| AFF1 IP recovers CDK9 in AFF4-null cells | AFF1 has an AFF4-independent route; hypothesis dead |
| AFF1 IP loses CDK9 in AFF4-null cells | consistent with AFF4-mediated recruitment; does not yet prove a direct AFF4-AFF1 contact |
| AFF1 IP loses CDK9 and AFF1 IP in parental cells recovers AFF4 | the AFF4-mediated topology, and it also converts the 8-spoke AFF1-AFF4 pair into a co-IP with a genetic control |
A co-IP is still an n-ary experiment, so arm B alone would produce exactly the kind of record this census grades weakest. What upgrades it is the genetic subtraction: losing the signal when a specific bridging candidate is deleted is topological information a pulldown cannot give on its own. Arm A remains the one that produces a binary measurement.
What this fixes in the graph regardless of the answer
This project currently asserts AFF1-CDK9, AFF1-CCNT1, AFF1-AFF4, AFF1-ELL2 and AFF1-MLLT1 on records that are 100% spoke-expanded 2. Whichever way the experiment goes, five interaction claims about the index fusion's own partner acquire a provenance better than curator inference - and a Kd, or a bound on one, is the kind of number the graph's interface entries are built from and currently cannot have for AFF1.
| design | Step 0: Europe PMC FULL-TEXT sweep for an existing binary AFF1/P-TEFb measurement, using the name list AFF1, AF4, MLLT2 against isothermal titration, surface plasmon, Kd, purified recombinant, GST pulldown. Arm A: recombinant AFF1 fragment (paralogous to the AFF4 segment in 4IMY) plus a large-fragment arm, titrated against purified CDK9/cyclin T1 by ITC and SPR, with the equivalent AFF4 fragment as positive control and a charge-matched scramble as negative; analytical SEC as a folding/monodispersity check. Arm B: AFF1 immunoprecipitation from parental and AFF4-knockout KMT2A::AFF1 cells (SEM or RS4;11) and a non-rearranged control, blotting CDK9, cyclin T1, ELL2, MLLT1 and AFF4; reciprocal AFF4 IP; AFF1-knockout lysate as antibody specificity control. |
|---|---|
| readout | Arm A: a dissociation constant, or an explicitly bounded negative stating the concentration ceiling reached (Kd > X). Arm B: presence or absence of CDK9 and cyclin T1 in the AFF1 IP as a function of AFF4 genotype. |
| controls | The AFF4 fragment must bind in the same session on the same P-TEFb preparation - an AFF1 negative without a same-session positive is uninterpretable. AFF1-knockout lysate for antibody specificity. A matched non-target gene set if any chromatin readout is added later. Report the Kd ceiling, never 'no interaction'. |
| cost tier | low |
| timeline | step 0 one day; arm B four to six weeks if AFF4-null cells exist, longer if they must be made; arm A six to ten weeks including construct scanning |
| what it cannot do | Arm B is itself an n-ary experiment and would generate exactly the class of record this census grades weakest; only arm A produces a binary measurement. Neither arm addresses whether the KMT2A::AFF1 fusion protein, as opposed to wild-type AFF1, makes the contact - that needs the fusion as the bait and is a separate design. |
tests hypotheses
[ "hyp-aff1-reaches-ptefb-only-through-aff4" ]
converts these spoke only pairs
[ "AFF1-CDK9", "AFF1-CCNT1", "AFF1-AFF4", "AFF1-MLLT1", "AFF1-ELL2" ]
calibration structures
[ "4IMY", "4OGR", "4OR5", "5L1Z", "6CYT" ]
- A clean negative in arm A would be read as 'AFF1 does not bind P-TEFb', when all it can support is 'no binding detectable between these constructs under these conditions' - and intrinsically disordered, multivalent, low-affinity contacts are exactly the class that purified-fragment biophysics misses. The bounded-Kd reporting requirement is not a formality here; it is the difference between a result and a mistake.
- If the answer is AFF4-mediated, the obvious next move is to target AFF4 - a general elongation factor required in normal haematopoiesis, so the therapeutic index would be worse than menin's, not better.
- Arm B's AFF4 knockout will reduce global elongation, so any fusion-target-specific interpretation needs a matched non-target normalisation or the blot will show a collapse that means nothing about topology.
- Which AFF1 isoform do t(4;11) lines actually express, and does the canonical UniProt numbering correspond to it? Construct design assumes it does and nothing here checked.
- Should the fusion protein itself be an arm? The KMT2A::AFF1 chimera retains a defined AFF1 region and the whole question is about the fusion, but a fusion construct is harder to purify and the wild-type answer comes first.
- Is there an AFF4-knockout leukaemia line already in circulation, or would one have to be made? The timeline turns on this and it was not checked.
- E1established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database
PDB 4IMY is an x-ray structure whose three polymer entities map to CDK9 (P50750), cyclin T1 (O60563) and AFF4 (Q9UHB7), providing the paralogous construct boundaries and the positive-control interaction for an AFF1 binding assay against purified P-TEFb.
The AFF4 scaffold binds human P-TEFb adjacent to HIV Tat, PDB 4IMY, entry and polymer-entity records with each entity's UniProt mapping read back individually, fetched 2026-09-21 · RCSB Protein Data Bank 2026What this evidence cannot show
The structure was determined in the context of HIV Tat transactivation, so the AFF4 construct boundaries were optimised for that complex and may not transfer cleanly to AFF1 by alignment alone; a short construct scan is likely to be needed and is budgeted. The residue ranges resolved in the entry were not extracted in this beat - only the entity identity and UniProt mapping were - so construct design requires one further read of the coordinates.
Evidence that argues against this (1)
- Designing an AFF1 construct by paralogy to an AFF4 structure builds the parsimonious hypothesis into the experiment. If AFF1 binds P-TEFb through a DIFFERENT segment, a paralogy-guided construct scan would return a false negative and would look exactly like the result the hypothesis predicts. The mitigation is to include a full-length or large-fragment AFF1 arm, not only the paralogous segment. RCSB Protein Data Bank 2026
- E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.database
In the IntAct REST record fetched 2026-09-21 every record for AFF1-CDK9 (16), AFF1-CCNT1 (8), AFF1-AFF4 (8), AFF1-MLLT1 (9) and AFF1-ELL2 (2) is MI:1060 spoke-expanded, with no MI:0407 record and no PDB entry containing both chains for any of the five pairs.
IntAct REST ws/interaction/findInteractions/P51825 joined to the P-TEFb, AFF4, MLLT1 and ELL2 accessions, fetched 2026-09-21; analyses/spoke-expansion-provenance-census/run.py · EMBL-EBI IntAct 2026What this evidence cannot show
Bounded by the curated record. Step 0 of this experiment exists precisely because IntAct does not index measurements that appear only in supporting figures, and the full-text sweep that would bound the literature rather than the database has NOT been done - so this is a reason to look, not an established absence.
Evidence that argues against this (1)
- These records come from focused complex purifications by labs that had characterised the SEC biochemically, not from atlases. Their reproducibility across tags, cells and labs is strong evidence of co-membership, and an experiment framed as filling an evidence gap should not be sold as correcting an error. Yokoyama A et al. 2010
- E3established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database
AFF1 is UniProt P51825, 1,210 aa, primary gene name AFF1, and AFF4 is Q9UHB7, 1,163 aa, primary gene name AFF4 - both read back from live UniProt - which are the construct-design starting points for the alignment this experiment depends on.
UniProtKB entries P51825 (AFF1_HUMAN) and Q9UHB7 (AFF4_HUMAN), accession, primary gene name, length and organism read back 2026-09-21 · UniProtKB 2026What this evidence cannot show
Accessions resolved by reading the PRIMARY gene name back rather than by trusting a symbol search, per this project's standing rule; both returned Homo sapiens and the expected symbol. AFF1's previous official symbol MLLT2 and the common name AF4 reach literature the current symbol does not, which matters for step 0's name list and is stated there.
Evidence that argues against this (1)
- Both proteins have multiple isoforms and the canonical UniProt sequence is not necessarily the one expressed in a leukaemic blast. A construct designed against the canonical numbering may not correspond to the dominant cellular species, and this beat did not check which isoform t(4;11) lines express. UniProtKB 2026
ITC and SPR against purified P-TEFb with AFF4 absent, plus AFF1 IP from AFF4-null cells
converts five spoke-only AFF1 pairs into a dissociation constant or a bounded negative
purified CDK9/cyclin T1 as the analyte, with the AFF4 fragment as the same-session positive control