Thomas Milne (MRC WIMM, Oxford) — MLL-AF4 target biology, enhancer architecture and PROM1/CD133
also called Milne Lab · MRC Molecular Haematology Unit Oxford · Weatherall Institute of Molecular Medicine
Milne's group in Oxford works out exactly which genes the MLL-AF4 fusion switches on and how it physically loops DNA to do it. Their discovery that the fusion drives CD133 has already become a CAR-T target.
The same thing, for a scientist
Thomas A. Milne runs a laboratory in the MRC Molecular Haematology Unit, MRC Weatherall Institute of Molecular Medicine, Radcliffe Department of Medicine, University of Oxford. The group's speciality is high-resolution chromatin architecture at MLL-AF4 target loci: they mapped an intragenic H3K79me2/3 enhancer element controlling PROM1/CD133, showed by Micro-Capture-C that MLL-AF4 binding creates broad high-density enhancer-promoter interaction domains enriched for PAF1C and FACT, and their PROM1 result has now been carried into a bispecific CD19-CD133 CAR-iNKT programme with the Karadimitris and Roy groups.
What they own that the Iqbal Lab does not1 deeper
Ultra-high-resolution chromatin conformation on primary and model KMT2A-r cells. Micro-Capture-C at base-pair resolution is a specialist technique with a narrow set of practitioners, and Milne's group applied it to show that MLL-AF4 binding promotes broad, high-density regions of enhancer-promoter interactions at a subset of key targets, enriched for elongation factors including PAF1C and FACT 2. A lab without that capability cannot make claims about fusion-driven 3D contacts and will be refereed by people who can.
A validated fusion-specific surface antigen. PROM1/CD133 is a direct MLL-AF4 target whose expression is essential for leukemic cell growth, controlled by an intragenic H3K79me2/3 enhancer element and coupled to the neighbouring TAPT1 locus 1. This is the rarest thing in the field: a cell-surface consequence of the fusion. It converts a chromatin finding into an immunotherapy target, and it is now in a bispecific CD19-CD133 CAR-iNKT construct that eradicated both medullary and leptomeningeal KMT2A-r leukaemia in models 3.
Deep integration with the Oxford paediatric haematology network. The PROM1 and CAR-iNKT papers pull in Anindita Roy and Irene Roberts (Oxford Paediatrics, fetal haematopoiesis), Great Ormond Street Hospital, Imperial College, and Glasgow 13. Milne is a node in a UK infant-leukaemia network, not a solo lab.
Why this is a natural collaborator rather than a competitor
Milne's output is target biology, not drug development. He publishes what the fusion does to chromatin and which genes it controls; he does not run a menin programme or a chemistry programme. That makes the overlap with an interface-disruption lab small and the complementarity large.
The concrete ask is well defined: if the Iqbal Lab predicts that disrupting a given KMT2A::AFF1 partner interface should collapse a specific target-gene enhancer contact, Milne's group has the only assay that can test that prediction directly. Conversely, the reciprocal offer is a computational interface map that tells them which partner to deplete next.
A second, cheaper angle: the PROM1/CD133 finding implies there may be other fusion-driven surface antigens that nobody has systematically enumerated. That enumeration is a bioinformatics project a small lab can run on public data, and it is directly useful to them.
The uncomfortable finding in their orbit
A joint Oxford-Princess Maxima paper modelled acquired resistance to the DOT1L inhibitor pinometostat in paediatric KMT2A::AFF1B ALL cells and found something that should worry anyone betting on a single mechanistic dependency 4. The resistant cells became largely independent of DOT1L-mediated H3K79 methylation yet still required the physical presence of DOT1L, HOXA9 and the KMT2A::AFF1 fusion. They selectively lost epigenetic regulation and expression of some fusion targets, including PROM1/CD133, while HOXA9 and CDK6 were unaffected, and they upregulated myeloid genes including CD33 and LILRB4.
Read that carefully. It says a fusion-driven leukaemia can drop the catalytic requirement while keeping the scaffold, can shed an individual fusion target gene without dying, and can drift toward a myeloid identity under targeted pressure. Every one of those is a failure mode for an interface-disruption strategy, and it also means the CD133 CAR target can be lost under drug pressure.
| group | Milne Laboratory (Thomas A. Milne); close collaborators Nicholas T. Crump (now Imperial College London), Anindita Roy and Irene Roberts (Oxford Paediatrics) |
|---|---|
| institution | MRC Molecular Haematology Unit, MRC Weatherall Institute of Molecular Medicine, Radcliffe Department of Medicine, University of Oxford, Oxford, UK (affiliation as printed on PubMed-indexed papers through 2026) |
| focus | MLL-AF4/KMT2A-AF4 direct target genes, H3K79 methylation, intragenic enhancers, base-pair-resolution enhancer-promoter contacts (Micro-Capture-C), PROM1/CD133 as a fusion-driven surface antigen |
| partnering angle | Highest-value natural collaborator. Offer a computational partner-interface map that predicts which KMT2A::AFF1 interaction to disrupt; ask for Micro-Capture-C readout of whether that disruption collapses specific enhancer-promoter contacts. Second, cheaper angle: systematically enumerate other fusion-driven cell-surface antigens beyond PROM1 from public KMT2A-r expression data, which is directly useful to their CAR programme and costs us nothing but analysis time. |
| relationship | natural collaborator, minimal overlap |
key papers — the numbers behind the chart
[
{
"title": "H3K79me2/3 controls enhancer-promoter interactions and activation of the pan-cancer stem cell marker PROM1/CD133 in MLL-AF4 leukemia cells",
"year": 2020,
"venue": "Leukemia",
"pmid": "32242051",
"doi": "10.1038/s41375-020-0808-y"
},
{
"title": "MLL-AF4 cooperates with PAF1 and FACT to drive high-density enhancer interactions in leukemia",
"year": 2023,
"venue": "Nature Communications",
"pmid": "37626123",
"doi": "10.1038/s41467-023-40981-9"
},
{
"title": "Modelling acquired resistance to DOT1L inhibition exhibits the adaptive potential of KMT2A-rearranged acute lymphoblastic leukemia",
"year": 2023,
"venue": "Exp Hematol Oncol",
"pmid": "37740239",
"doi": "10.1186/s40164-023-00445-8"
},
{
"title": "Off-the-shelf dual CAR-iNKT cell immunotherapy eradicates medullary and leptomeningeal high-risk KMT2A-rearranged leukemia",
"year": 2026,
"venue": "Blood",
"pmid": "40898981",
"doi": "10.1182/blood.2025029302"
},
{
"title": "The MLL family of proteins in normal development and disease (review)",
"year": 2020,
"venue": "BBA Gene Regul Mech",
"pmid": "33157274",
"doi": "10.1016/j.bbagrm.2020.194648"
}
]leverage they hold
[ "Micro-Capture-C base-pair-resolution 3C capability", "validated MLL-AF4 direct-target gene set with enhancer annotation", "PROM1/CD133 immunotherapy translation route through Imperial and Great Ormond Street", "UK infant-leukaemia network (Oxford Paediatrics, GOSH, Glasgow, Imperial)" ]
- Their own resistance model shows a KMT2A-r cell can keep the fusion and the scaffold while dropping a catalytic dependency and shedding individual target genes, which is a general failure mode for single-interface disruption.
- CD133 is a pan-cancer stem cell marker expressed on normal progenitors, so a CD133-directed therapy carries on-target off-tumour risk that murine xenografts cannot detect.
- Micro-Capture-C claims are hard to independently replicate, so building a hypothesis on their 3D-contact model creates a dependency on a single group's technique.
- Is PROM1/CD133 the only fusion-driven surface antigen, or is there an unenumerated set? A systematic surfaceome analysis of KMT2A-r expression data has not been published and is within reach of a small lab.
- Does disrupting the menin interface collapse the high-density enhancer-promoter contact domains that Micro-Capture-C detects, or do those contacts persist? Nobody appears to have run menin inhibition through this assay, and the answer decides whether menin inhibition is architectural or merely transcriptional.
- If pinometostat-resistant cells lose PROM1/CD133 while keeping the fusion, will menin-inhibitor-resistant cells lose it too, and does that pre-emptively sabotage a CD133 CAR given after a menin inhibitor?
- E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
Milne's group showed PROM1/CD133 is a direct MLL-AF4 target whose aberrant expression is essential for leukemic cell growth, activated through an intragenic H3K79me2/3 enhancer element that drives enhancer-promoter interactions between PROM1 and the neighbouring TAPT1 gene, with PRC2 repression in non-expressing cells.
H3K79me2/3 controls enhancer-promoter interactions and activation of the pan-cancer stem cell marker PROM1/CD133 in MLL-AF4 leukemia cells · Godfrey L, Crump NT, O'Byrne S, et al., Roy A, Milne TA · Leukemia 2020What this evidence cannot show
Largely cell-line based (SEM and related MLL-AF4 models) with primary-sample validation; 'essential for leukemic cell growth' derives from knockdown in those models rather than from in-vivo genetic ablation in primary infant disease. Pablo Menendez is a co-author, so the Oxford and Barcelona groups are already linked.
Evidence that argues against this (1)
- This entry's own E4 contradicts the word 'essential' in E1. In the joint Oxford-Princess Maxima model of acquired pinometostat resistance, paediatric KMT2A::AFF1 ALL cells SELECTIVELY LOST PROM1/CD133 expression and continued to grow, which is not compatible with CD133 expression being required for leukemic cell growth. The reconciling reading is that CD133 is required in the DOT1L-dependent state and dispensable once the cell rewires away from H3K79 methylation - which, if right, predicts that CD133-directed therapy selects for exactly the adaptive state that DOT1L inhibitors also select for. Schneider P et al. 2023
- E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
Using Micro-Capture-C, the group showed MLL-AF4 binding promotes broad, high-density regions of enhancer-promoter interactions at a subset of key target genes, enriched for the transcription elongation factors PAF1C and FACT.
MLL-AF4 cooperates with PAF1 and FACT to drive high-density enhancer interactions in leukemia · Crump NT, ... Milne TA · Nature Communications 2023What this evidence cannot show
Micro-Capture-C is a specialist technique with few independent replications outside Oxford; the causal direction (fusion binding creates the contacts, versus pre-existing contacts recruit the fusion) is inferred from perturbation rather than proven. The finding that co-activators such as BRD4 and Mediator have little impact on enhancer-promoter interactions is itself contested in the wider 3D-genome literature.
Evidence that argues against this (1)
- Micro-Capture-C reports CONTACT FREQUENCY averaged over a population of cells, so broad high-density interaction regions describe an ensemble and not a structure any one nucleus holds. The PAF1C and FACT enrichment is computed by overlap with existing occupancy data, which inherits those datasets' antibodies and thresholds. eLife 2021
- E3moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper
The PROM1/CD133 finding has been translated into an off-the-shelf bispecific CD19-CD133 CAR-iNKT cell therapy that eradicated medullary and leptomeningeal KMT2A-rearranged leukaemia in models, with Milne and Anindita Roy as senior co-authors.
Off-the-shelf dual CAR-iNKT cell immunotherapy eradicates medullary and leptomeningeal high-risk KMT2A-rearranged leukemia · Ren H, Elliott N, Lye B, et al., Milne TA, Roy A, Karadimitris A · Blood 2026What this evidence cannot show
Preclinical in-vitro and xenograft data only; 'no discernible haematologic toxicity' in immunodeficient mice says little about human on-target off-tumour risk, and CD133 is expressed on normal stem and progenitor populations in several tissues. Published 2026-01-08 with a 2026 volume, so it is very recent and unreplicated.
Evidence that argues against this (1)
- CD133 is not a leukemia-restricted antigen, so 'aberrant expression' understates the on-target risk of a CD133-directed cell therapy. Prominin-1/CD133 is a canonical marker of normal human haematopoietic stem and progenitor cells and is additionally expressed on adult human differentiated epithelia of kidney, and on certain kidney cancers - a distribution incompatible with the assumption of a clean tumour-specific target. A dual CD19-CD133 CAR is therefore adding a second antigen whose normal-tissue expression profile is broader than CD19's, not narrower. Florek M et al. 2005
- E4moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper
A joint Oxford-Princess Maxima model of acquired pinometostat (DOT1L inhibitor) resistance in paediatric KMT2A::AFF1B ALL cells became largely independent of H3K79 methylation while still requiring the physical presence of DOT1L, HOXA9 and the fusion, selectively lost PROM1/CD133 expression, and upregulated myeloid genes including CD33 and LILRB4.
Modelling acquired resistance to DOT1L inhibition exhibits the adaptive potential of KMT2A-rearranged acute lymphoblastic leukemia · Schneider P, Crump NT, Arentsen-Peters STCJM, et al., Milne TA, Pieters R, Stam RW · Experimental Hematology & Oncology 2023What this evidence cannot show
In-vitro selected resistance in a single cell-line background; selected resistance phenotypes frequently do not match what emerges in patients, and pinometostat's clinical failure had pharmacological as well as biological causes. The paper is nonetheless a direct demonstration that a KMT2A-r cell can uncouple catalytic dependency from scaffold dependency.
Evidence that argues against this (1)
- Acquired resistance derived in cultured cells under prolonged drug exposure is a strong and unphysiological selection, and the finding — independence from H3K79 methylation while still requiring DOT1L physically — is a property of the surviving line. It aligns with the published non-canonical H3K79me2-dependent survival pathways, which is the corroboration worth citing here. eLife 2021
- ← proposed byStep 5 — The output program: HOXA9/HOXA10/MEIS1/PBX3 plus MEF2C, MYB, BCL2, FLT3 and PROM1/CD133
MLL-AF4 target biology and enhancer usage
- related to →KMT2A::AFF1 (MLL-AF4), t(4;11)(q21;q23)
maps MLL-AF4 direct target genes and the enhancer-promoter architecture they sit in
H3K79me2/3 enhancer element (KEE) controls PROM1/CD133 activation; also co-authored the pinometostat resistance model
priority-three natural collaborator: unique assay, near-zero competitive overlap
- related to →Pinometostat (EPZ-5676) — the DOT1L inhibitor, and the most instructive failure in this field
co-authored the acquired-resistance model showing KMT2A-r cells uncouple catalytic H3K79 dependency from scaffold dependency
priority collaboration 3
Menendez is a co-author on the Oxford PROM1/CD133 paper; the Oxford-Barcelona axis already exists
joint Oxford-Princess Maxima pinometostat resistance model (Milne, Pieters, Stam)
PROM1/CD133 is the worked example of turning a fusion-driven target into an immunotherapy antigen