Oncofusion

start / graph / gene-aff2

Proteinmoderate confidence 3 of 5: One strong study, or several smaller consistent ones.Q5d-interactome3

AFF2 (FMR2) — the SEC-L2 scaffold, and the AFF-family member with no dependency anywhere

also called AFF2 · FMR2 · FMR2P · OX19 · AF4/FMR2 family member 2

AFF2 is the fourth member of the family of scaffold proteins that build the cell's transcription-speeding machine. Unlike its relatives it is switched off in most cells and no cell line tested needs it, so it is the family member that tells you the family slot can be empty.

The same thing, for a scientist

AFF2 (P51816, 1,311 aa) assembles **SEC-L2**, a super-elongation-complex-like assembly containing P-TEFb with ENL/MLLT1 and AF9/MLLT3, isolated in the same study that defined SEC-L3. SEC-L2 controls a gene subset distinct from SEC and SEC-L3 and is not required for stress induction of HSP70. AFF2 is not a component of KMT2A::AFF1 and does not bind it; it is entered here to complete the AFF-scaffold family, because the family is the correct unit for any specificity argument about the fusion's 3' moiety. Its dependency profile is the family's null: mean gene effect **+0.105** across 1,208 DepMap lines with **0.17%** of lines depleted and a KMT2A-rearranged minus other-leukemia difference of **-0.006**. *AFF2* is also the gene silenced in FRAXE intellectual disability, which is the only setting in which its loss has a documented human phenotype.

What it looks like
loading AF-P51816-F1…
AlphaFold AF-P51816-F1 ↗
colourshape

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.

very high (90+)confident (70–90)low (50–70)very low (<50)
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.

Go deeper
What SEC-L2 is, and what its existence establishes

SEC-L2 was isolated biochemically alongside SEC-L3, containing AFF2 in association with P-TEFb, ENL/MLLT1 and AF9/MLLT3 1. Like the other SEC species it has high Pol II CTD kinase activity; unlike AFF4/SEC it is not required for proper HSP70 induction on stress, and genome-wide mRNA-seq showed it controls a different subset of genes 1.

The argument this supports is a negative one about specificity. The ENL and AF9 contacts are not properties of AFF1 — they are properties of the AFF family, present in SEC, SEC-L2 and SEC-L3 alike. Anything that engages ENL or AF9 engages all of them.

Method: complex isolation with mass-spectrometric composition, CTD kinase assay, RNAi plus mRNA-seq 1. Shown for: wild-type complexes in HEK293 and Jurkat. Which part of the fusion: none. AFF2 is a paralogue of the AFF1 moiety, not a binder of the fusion. AFF2 is not a member of the KMT2A::AFF1 interactome.

The dependency null, and why it is worth recording

AFF2 mean gene effect is +0.105 across 1,208 lines, 0.17% of lines depleted, with a KMT2A-rearranged minus other-leukemia difference of -0.006 2. That is as close to nothing as this project's data produce.

Put beside its relatives, the AFF family spans almost the whole range of dependency behaviour: AFF4 and AFF1 carry the fusion biology, AFF3 shows a small negative difference (-0.085), and AFF2 shows nothing at all. A family whose members range from indispensable to invisible is a family in which membership alone predicts nothing, and that is the correct caution to apply whenever 'the AFF scaffold' is treated as a single entity in a figure.

The one human phenotype attached to AFF2 loss is FRAXE — a CCG-repeat expansion and methylation silencing at AFF2 causing mild intellectual disability. That it is a neurodevelopmental rather than haematopoietic phenotype is itself informative about where this scaffold matters.

Facts and figures
uniprotP51816
gene symbolAFF2
length AA1311
protein nameAF4/FMR2 family member 2
alphafold IDAF-P51816-F1
accession resolutionUniProt query gene_exact:AFF2 AND organism_id:9606 AND reviewed:true; primary gene name read back as AFF2 and length 1311 confirmed
complexSEC-L2 (super elongation complex-like 2): AFF2 + P-TEFb + ENL/MLLT1 + AF9/MLLT3
human loss of function phenotypeFRAXE intellectual disability (CCG repeat expansion and methylation silencing of AFF2); neurodevelopmental, not haematopoietic
interaction evidence classbiochemical complex isolation with MS composition, CTD kinase assay, RNAi + mRNA-seq
shown for fusionfalse
contact on kmt2a aff1none — paralogue of the AFF1 moiety, not a binder
druggabilityNot a target. Entered to complete the AFF family and to anchor the specificity argument.
aff family
[
  "AFF1 (P51825)",
  "AFF2 (P51816)",
  "AFF3 (P51826)",
  "AFF4 (Q9UHB7)"
]
essentiality
{
  "depmap_common_essential": false,
  "chronos_mean": 0.1052,
  "n_lines": 1208,
  "fraction_lines_depleted": 0.0017,
  "chronos_mean_kmt2a_rearranged": 0.0225,
  "chronos_mean_other_leukemia": 0.0282,
  "kmt2a_selectivity_difference": -0.0057,
  "n_kmt2a_rearranged_lines": 11,
  "n_other_leukemia_lines": 43,
  "benchmark_men1_difference": -0.8656
}
How this could bite us back
  • An AFF-scaffold-directed agent cannot rely on AFF2's dispensability as a safety argument, because AFF2 loss has a documented human neurodevelopmental phenotype that cell-line screens cannot see.
  • Paralogue redundancy across the AFF family is the main structural reason a SEC-scaffold strategy should be expected to be buffered.
Still open
  1. Is AFF2 expressed at all in fetal-liver or pro-B progenitors? If not, SEC-L2 is irrelevant to this disease and the AFF family effectively has three members here rather than four.
  2. Do the four AFF scaffolds compete for a limiting pool of ENL and AF9, and does removing one raise the availability of those partners to the fusion?
  3. AFF2 silencing causes a neurodevelopmental syndrome and no blood phenotype. Does that tissue restriction hold at the protein level in haematopoiesis, or is it simply unmeasured?
Every source behind this entry (2)
  1. E1established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper

    SEC-like 2 (SEC-L2), containing AFF2 in association with P-TEFb, ENL/MLLT1 and AF9/MLLT3, was biochemically isolated; it has high Pol II CTD kinase activity but is not required for proper HSP70 induction upon stress; and genome-wide mRNA-seq showed SEC-L2 and SEC-L3 control different subsets of genes from AFF4/SEC.

    The super elongation complex family of RNA polymerase II elongation factors: gene target specificity and transcriptional output. · Luo Z, Lin C, Guest E, Garrett AS, Mohaghegh N, Swanson S, Marshall S, Florens L, Washburn MP, Shilatifard A · Molecular and Cellular Biology 2012
    PMID 22547686DOI
    What this evidence cannot show

    Abstract read in full; not full text. HEK293 and Jurkat cells; no KMT2A fusion protein tested. AFF2 is expressed narrowly in normal tissue, so its presence in a complex isolated from these lines does not establish that SEC-L2 exists in a B-lymphoid progenitor.

    Evidence that argues against this (1)
    • 'AFF2 is not a component of KMT2A::AFF1 and does not bind it' is stated flatly and rests on absence of evidence, and there is a specific published mechanism by which it could be false. Chen and Cramer solved the AFF4 C-terminal homology domain (CHD) at 2.2 A and show it 'mediates the formation of an AFF4 homodimer or an AFF1-AFF4 heterodimer'; the CHD is conserved across AFF1, AFF2, AFF3 and AFF4. In AFF1 the CHD lies at aa 949-1209 - downstream of every observed KMT2A::AFF1 breakpoint and therefore fully retained in the fusion (this graph's own hyp-aff1-chd-dimerisation-in-kmt2a-fusion). A CHD-mediated AFF2-AFF1 heterodimer with the fusion is therefore a structurally licensed, untested possibility. The honest status of this claim is 'not tested', not 'does not bind'. Read: abstract in full, not full text. Chen Y et al. 2019
  2. E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.dataset

    AFF2 has DepMap mean gene effect +0.1052 across 1,208 lines with 0.17% of lines depleted, mean +0.0225 in 11 KMT2A-rearranged leukemia lines against +0.0282 in 43 other leukemia lines (difference -0.006); MEN1 in the same lines is -0.866 and AFF3 is -0.085.

    DepMap CRISPR (Chronos) gene effect via the Open Targets Platform GraphQL API, with KMT2A-rearrangement status from Cellosaurus; per-gene recomputation for this beat · Broad Institute DepMap; Cellosaurus; Open Targets Platform; analysis run by this project · Database and derived analysis 2026
    What this evidence cannot show

    DepMap release version not stated by the API. A gene expressed in few of the screened lines scores as non-essential regardless of its biology, and AFF2's narrow expression makes that the most likely explanation for this null rather than a statement about function.

    Evidence that argues against this (1)
    • The 'no dependency anywhere' reading (+0.105 mean, 0.17% of lines depleted) is the profile a CRISPR screen returns for a scaffold whose function is buffered by three paralogues that all carry the same dimerisation and P-TEFb-binding modules. Chen and Cramer's demonstration that the AFF CHD is a general homo- and heterodimerisation module across AFF1-4 supplies the mechanism for exactly that buffering. A flat single-gene knockout profile in a four-member scaffold family constrains nothing about whether the family matters; the informative experiment is combinatorial, and this graph already carries hyp-paralogue-buffering-hides-dependencies for the general case. Read: abstract in full, not full text. Chen Y et al. 2019
Connected to