KDM1A (LSD1) — acts at the loci the fusion binds, spares normal stem cells, and still has never been shown to touch the fusion protein
also called KDM1A · LSD1 · AOF2 · BHC110 · KDM1 · Lysine-specific histone demethylase 1A
LSD1 is an enzyme that strips activating marks off histones. In MLL-AF9 leukaemia it sits at exactly the places the fusion binds and keeps the oncogenic programme switched on, and blocking it lets the cells differentiate while leaving normal blood stem cells alone — one of the cleanest therapeutic windows in this whole field.
The same thing, for a scientist
KDM1A (O60341, 852 aa) is the FAD-dependent H3K4me1/2 demethylase of the CoREST complex, with a SWIRM domain at aa 174-273 and two long coiled coils at aa 110-151 and 428-514 [E1]. In a mouse model of human MLL-AF9 leukaemia it was identified as an **essential regulator of leukaemia stem cell potential**, acting **at genomic loci bound by MLL-AF9** to sustain the associated oncogenic programme and prevent differentiation and apoptosis; pharmacological targeting with nanomolar tranylcypromine analogues phenocopied Kdm1a knockdown in murine and primary human AML with MLL translocations, while **the clonogenic and repopulating potential of normal haematopoietic stem and progenitor cells was spared** [E2]. The dependency data agree in direction and are modest in size: KDM1A ranks **331 of 17,728 (1.867th percentile)** with a difference of **-0.1762** on a pan-essentiality of -0.2233 [E3]. The graph already carries the clinical side as `target-lsd1-kdm1a-kmt2a-r` and the GFI1 contact as `interface-lsd1-gfi1-snag`; this entry supplies the protein and the interactome classification. **Classification: dependency at fusion-bound loci, not a demonstrated fusion interactor** — co-occupancy on chromatin is not a protein-protein interaction, and no co-IP with any KMT2A fusion is cited here.
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.
'Acts at loci bound by MLL-AF9' is the strongest chromatin-level statement in this beat, and it is still not a binding result
Most entries in this beat had to settle for 'the gene is required' or 'the complex is in the same pathway'. Harris et al. showed something considerably better: KDM1A acts at the genomic loci bound by MLL-AF9, sustaining expression of the fusion's programme 2. That is co-occupancy plus functional dependence at the same sites.
It is still not evidence that KDM1A contacts the fusion protein. Three mechanisms fit the data equally well: KDM1A is recruited by the fusion; KDM1A is recruited independently by a transcription factor at the same enhancers (GFI1 is the obvious candidate, and the graph carries interface-lsd1-gfi1-snag); or KDM1A maintains a chromatin state the fusion requires without either protein ever meeting. The entry is filed as a dependency for that reason.
The practical difference matters. If KDM1A is recruited by the fusion, a menin inhibitor that displaces the fusion should also displace KDM1A, and the two agents would be redundant. If KDM1A arrives independently, they are complementary. Nobody has run the experiment.
The therapeutic window, and what it costs
The 2012 result includes the thing this field almost never has: an explicit normal-tissue control that came out clean. Tranylcypromine-analogue KDM1A inhibitors active in the nanomolar range phenocopied Kdm1a knockdown in murine and primary human MLL-translocated AML, while sparing the clonogenic and repopulating potential of normal HSPC 2. The DepMap pan-essentiality of -0.223 is consistent — KDM1A is not a gene every cell needs 3.
What it costs is in target-lsd1-kdm1a-kmt2a-r and in bite_back below: LSD1 inhibitors are differentiation agents, the clinical experience is phase 1/2a with no randomised evidence and no paediatric data, and the mechanism that makes them work in MLL-rearranged AML is the mechanism that produces differentiation syndrome.
Cell type, cell state, localisation, epigenetic readout
| Axis | Documented | Where |
|---|---|---|
| Cell type | Mouse MLL-AF9 leukaemia; primary human AML with MLL translocations; normal human HSPC as the sparing control | 2 |
| Cell state | Leukaemia stem cell potential is the readout; KDM1A loss permits differentiation and apoptosis | 2 |
| Subcellular | Nucleus | 1 |
| Subnuclear | Not established in a KMT2A-rearranged cell | - |
| Chromatin position | At genomic loci bound by MLL-AF9 - the most specific chromatin localisation any entry in this beat can claim | 2 |
| Epigenetic readout | H3K4me1 and H3K4me2 demethylation (FAD-dependent); also scaffolds CoREST and GFI1-containing repressive complexes | 14 |
| Lineage | Myeloid. No KMT2A::AFF1 or B-lineage result, and no paediatric data. | 2 |
| uniprot | O60341 |
|---|---|
| gene symbol | KDM1A |
| protein name | Lysine-specific histone demethylase 1A |
| length AA | 852 |
| alphafold ID | AF-O60341-F1 |
| accession resolution | UniProt REST query gene_exact:KDM1A AND organism_id:9606 AND reviewed:true, run 2026-09-19; one reviewed hit; primary gene name read back as KDM1A and length 852 confirmed from the same fetched record. |
| complex | CoREST; GFI1-containing repressive complexes |
| domains note | UniProt does not annotate a discrete amine-oxidase domain feature in the classes queried; no catalytic-domain coordinate is asserted rather than one being invented. |
| catalytic activity | FAD-dependent demethylation of H3K4me1 and H3K4me2 |
| classification | dependency acting at fusion-bound loci; NOT a demonstrated interactor of KMT2A::AFF1 |
| classification reason | the primary result is functional dependence and co-occupancy at MLL-AF9-bound loci, not a protein-protein interaction with the fusion; an independent recruitment route via the GFI1 SNAG domain is documented in this graph |
| interaction evidence class | mouse model genetics; chromatin co-occupancy at fusion-bound loci; pharmacological phenocopy; unbiased DepMap scan |
| shown for fusion | MLL-AF9 in mouse; primary human AML with MLL translocations; nothing in KMT2A::AFF1 or B-lineage |
| epigenetic readout | loss of H3K4me1/me2 demethylation on inhibition; sustained expression of the MLL-AF9 oncogenic programme when active |
| normal tissue control | clonogenic and repopulating potential of normal haematopoietic stem and progenitor cells spared by nanomolar tranylcypromine analogues (PMID 22464800) |
aliases retired or alternate
[ "LSD1", "AOF2", "BHC110", "KDM1" ]
domains — the numbers behind the chart
[
{
"name": "Coiled coil",
"start": 110,
"end": 151,
"source": "UniProt O60341, fetched 2026-09-19"
},
{
"name": "SWIRM",
"start": 174,
"end": 273,
"source": "UniProt O60341, fetched 2026-09-19"
},
{
"name": "Coiled coil",
"start": 428,
"end": 514,
"source": "UniProt O60341, fetched 2026-09-19"
}
]localization
{
"subcellular": "nucleus",
"subnuclear": "not established",
"chromatin_targeting": "genomic loci bound by MLL-AF9"
}related entries
[ "target-lsd1-kdm1a-kmt2a-r", "interface-lsd1-gfi1-snag", "gene-gfi1" ]
essentiality
{
"depmap_scan": "analyses/depmap-kmt2a-selective-scan-full",
"pan_essentiality_all_lines": -0.2233,
"n_all_lines": 1208,
"mean_kmt2a_rearranged": -0.4316,
"mean_other_leukemia": -0.2555,
"kmt2a_selectivity_difference": -0.1762,
"rank": 331,
"n_scanned": 17728,
"percentile": 1.867,
"interpretation": "top 2% selectivity with a shallow pan-essentiality - the profile a therapeutic window requires, though the absolute requirement is modest"
}enumeration record count
{
"query": "(MLL OR KMT2A) AND (KDM1A OR LSD1 OR AOF2 OR BHC110)",
"records": 41,
"date": "2026-09-19"
}depmap kmt2a selectivity
{
"chronos_mean_kmt2a_rearranged": -0.4316,
"chronos_mean_other_leukaemia": -0.2555,
"difference": -0.1762,
"pan_essentiality_all_lines": -0.2233,
"selectivity_rank": 331,
"rank_denominator": 17728,
"provenance": "Joined 2026-09-21 from analyses/depmap-kmt2a-selective-scan-full (generated 2026-09-19T10:12:06Z), the project's own genome-wide scan over 11 KMT2A-rearranged and 43 other leukaemia DepMap lines. NOT re-fetched and NOT independent of that scan: every caveat it carries applies here, including that the KMT2A-rearranged arm is a small number of lines drawn from nine individuals, that the rank denominator is 17,728 genes, and that a difference is not a dependency. Added because only 26 of 308 gene entries carried this number, which made any KMT2A-selectivity question answerable for 8% of the corpus.",
"how_to_read": "A negative difference means MORE required in KMT2A-rearranged lines than in other leukaemia lines. A rank near 17,728 is ANTI-selective. Neither is evidence of a fusion-driven dependency on its own: check pan_essentiality_all_lines before calling anything selective, and note that a paralogue-redundant gene can score near zero for reasons that have nothing to do with this disease - see ana-the-paralogue-redundancy-excuse-is-not-tested-and-my-proxy-decided-the-answer."
}- LSD1 inhibitors are differentiation agents in this disease, and induced differentiation in a KMT2A-rearranged patient is the same physiology that produces differentiation syndrome - the dose-limiting toxicity of the menin inhibitors this project is built around.
- Tranylcypromine-derived inhibitors carry monoamine oxidase liability, and the analogues in the founding paper are described by potency and not by MAO selectivity.
- KDM1A is required for normal erythroid and granulocytic maturation; the sparing of normal HSPC clonogenicity in one study is not the same as haematological safety in a treated infant.
- If KDM1A is recruited by GFI1 rather than by the fusion, then its requirement is a property of myeloid differentiation control rather than of the fusion, and selectivity in a B-lineage disease would have to be re-established from scratch.
- The clinical evidence is phase 1 and phase 2a with no randomisation and no paediatric data, so the window seen preclinically has not been shown to exist in patients.
- Does KDM1A contact a KMT2A fusion protein, or does it arrive at those loci independently, most likely via GFI1? A co-IP and a GFI1-null recruitment experiment would settle whether LSD1 inhibition and menin inhibition are redundant or complementary.
- Does menin inhibition displace KDM1A from fusion-bound loci? If it does, combining the two agents buys less than the mechanism suggests.
- Does KDM1A inhibition work in KMT2A::AFF1 infant B-ALL? Every published result in this disease is myeloid and adult, and the graph's clinical entry records no paediatric data.
- How much of the MLL-AF9 phenotype comes from demethylase activity and how much from CoREST/GFI1 scaffolding? Iadademstat inhibits both, so the clinical data cannot separate them.
- E1established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database
UniProtKB O60341 (KDM1A_HUMAN, Lysine-specific histone demethylase 1A) is 852 aa and annotates a SWIRM domain at aa 174-273 and coiled coils at aa 110-151 and aa 428-514.
UniProtKB O60341 (KDM1A_HUMAN), Lysine-specific histone demethylase 1A · UniProt Knowledgebase, reviewed (Swiss-Prot) 2026What this evidence cannot show
Record fetched 2026-09-19; primary gene name read back as KDM1A and length 852 confirmed from the same fetched record, and both coordinates copied from it. UniProt annotates no separate 'amine oxidase' domain feature in the feature classes queried here, so the catalytic region is not given a coordinate in this entry rather than one being invented.
Evidence that argues against this (1)
- The domain list omits the feature that matters most therapeutically. The tranylcypromine-analogue inhibitors act on the FAD-dependent amine oxidase activity, and the SWIRM domain and coiled coils describe the scaffolding functions instead - and iadademstat is reported to inhibit both the enzymatic and the scaffolding function, which a domain map alone cannot represent. Harris WJ et al. 2012
- E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
Using a mouse model of human MLL-AF9 leukaemia, KDM1A (LSD1, AOF2) was identified as an essential regulator of leukaemia stem cell potential; KDM1A acts at genomic loci bound by MLL-AF9 to sustain expression of the associated oncogenic programme, thus preventing differentiation and apoptosis; in vitro and in vivo pharmacological targeting with tranylcypromine analogues active in the nanomolar range phenocopied Kdm1a knockdown in both murine and primary human AML cells with MLL translocations; and by contrast the clonogenic and repopulating potential of normal haematopoietic stem and progenitor cells was spared.
The histone demethylase KDM1A sustains the oncogenic potential of MLL-AF9 leukemia stem cells · Harris WJ, Huang X, Lynch JT, Spencer GJ, Hitchin JR, Li Y, Ciceri F, Blaser JG, Greystoke BF, Jordan AM, Miller CJ, Ogilvie DJ, Somervaille TCP · Cancer Cell 21(4):473-87 2012What this evidence cannot show
Abstract fetched and read on 2026-09-19; full text not read. The fusion is **MLL-AF9**, myeloid; there is no KMT2A::AFF1 and no B-lineage experiment, and no paediatric material. 'Acts at genomic loci bound by MLL-AF9' is a co-occupancy-plus-function statement and is **not** a demonstration that KDM1A binds the fusion protein. Tranylcypromine analogues are mechanism-based MAO inhibitors and their selectivity against monoamine oxidases A and B is not stated in the abstract.
Evidence that argues against this (1)
- KDM1A also has a documented, fusion-independent recruitment route in this disease: it is held by the SNAG domain of GFI1, a transcription factor that operates at myeloid differentiation genes regardless of whether a KMT2A fusion is present. If the MLL-AF9-locus dependency runs through GFI1 rather than through the fusion, then KDM1A inhibition and menin inhibition are acting on different recruitment systems, and the strongest reading of this paper - that KDM1A serves the fusion - is unsupported. Salamero O et al. 2020
- E3moderate confidence 3 of 5: One strong study, or several smaller consistent ones.database
In the project's genome-wide KMT2A-rearranged selectivity scan of 17,728 genes, KDM1A ranks 331 (1.867th percentile) with a difference of -0.1762 (KMT2A-rearranged mean -0.4316 against other-leukaemia mean -0.2555) and a pan-essentiality of -0.2233 across 1,208 lines. For comparison in the same scan EP300 ranks 108 (-0.2837), DOT1L ranks 184 (-0.2310) and MEN1 ranks 1 (-0.8656).
Genome-wide KMT2A-rearranged selectivity scan, 17,728 genes (DepMap Chronos via Open Targets) · Broad Institute DepMap / Open Targets Platform; analyses/depmap-kmt2a-selective-scan-full 2026What this evidence cannot show
Rows read out of analyses/depmap-kmt2a-selective-scan-full/results.json on 2026-09-19. Eleven KMT2A-rearranged against 43 other leukaemia lines, AML-heavy, none infant-derived. Not permutation-tested for KDM1A. A shallow pan-essentiality of -0.223 is favourable for a therapeutic window but also means the absolute requirement in rearranged lines (-0.432) is modest.
Evidence that argues against this (1)
- The proliferation assay cannot see the phenotype the primary paper reports. Harris et al. measured leukaemia stem cell potential and repopulating capacity in vivo, not bulk growth in culture, and a factor required for self-renewal rather than proliferation would score weakly here. The agreement in direction is reassuring but the magnitude should not be taken as the size of the biological effect. Harris WJ et al. 2012
- E4moderate confidence 3 of 5: One strong study, or several smaller consistent ones.database
PubMed returns 41 records for '(MLL OR KMT2A) AND (KDM1A OR LSD1 OR AOF2 OR BHC110)' as of 2026-09-19, making KDM1A one of only three entries in this beat with a substantial literature in this disease (the others being EP300 and MYB).
PubMed E-utilities esearch record count for (MLL OR KMT2A) AND (KDM1A OR LSD1 OR AOF2 OR BHC110), run for this beat · NCBI PubMed, queried 2026-09-19 2026What this evidence cannot show
Symbol set searched: KDM1A, LSD1, AOF2, BHC110. Not searched: KDM1 (the pre-split symbol) as a standalone term, which would also match KDM1B. Only one of the 41 records was read in abstract for this entry, so the count establishes that a literature exists and nothing about its content.
Evidence that argues against this (1)
- A literature of 41 records is small enough that the field's view of KDM1A in this disease rests on a handful of studies, several of them from the same groups, and large enough that this entry's single-paper basis is not a fair summary of it. A future pass should read more of it before the entry's confidence is raised. Harris WJ et al. 2012
- depends on →KMT2A::MLLT3 (MLL-AF9), t(9;11)(p21.3;q23.3)
KDM1A acts at genomic loci bound by MLL-AF9 to sustain the oncogenic programme; nanomolar tranylcypromine analogues phenocopy knockdown while sparing normal HSPC
the protein and dependency evidence behind the clinical target entry
KDM1A is held by the GFI1 SNAG domain - a fusion-independent recruitment route that could explain the MLL-AF9-locus dependency without any fusion contact
- related to →KMT2A::AFF1 (MLL-AF4), t(4;11)(q21;q23)
NOT a demonstrated interactor: co-occupancy at fusion-bound loci plus functional dependence is not a protein-protein interaction
LSD1 inhibitor trial stops
LSD1 demethylates H3K4me1/me2 and not me3; inhibition raises both in KMT2A::AFF1 and KMT2A::MLLT3 lines
- depends on →KMT2A::MLLT3 (MLL-AF9), t(9;11)(p21.3;q23.3)
KDM1A acts at MLL-AF9-bound loci to sustain the oncogenic programme and LSC potential