Multipotent progenitor (MPP) — the stage where the undifferentiated KMT2A-r infant leukaemias map
also called MPP · haematopoietic multipotent progenitor cell · CD34+CD38-CD45RA-CD90- MPP
The multipotent progenitor is the step just after a blood stem cell: it can still make every kind of blood cell but can no longer renew itself indefinitely. Some of the most treatment-resistant infant KMT2A leukaemias look transcriptionally like this very early stage.
The same thing, for a scientist
The MPP (CL:0000837; marker-explicit CL:0002043) is the CD34-positive CD38-negative CD45RA-negative CD90-negative compartment immediately downstream of the HSC, retaining full lineage potential but with limited self-renewal. Its importance for this beat is empirical rather than theoretical: when 84 infant KMT2A-rearranged leukaemias were mapped onto haematopoietic differentiation by single-sample and single-cell inference, they distributed across defined stages of B-lymphopoietic and early hemato-endothelial development, and the most refractory integrative cluster was an IRX-type undifferentiated group. Differentiation stage, not fusion partner alone, tracked with prognosis.
The differentiation-stage map of infant KMT2A-r leukaemia1 deeper
A multi-omics study of 84 infants with KMT2A-rearranged leukaemia combined RNA sequencing, methylation arrays, whole-exome and targeted deep sequencing, then used single-sample and single-cell inference of haematopoietic differentiation to place each case on a developmental axis. It resolved five robust integrative clusters (ICs), each with distinct master transcription factors, fusion partners and corresponding developmental stages 1:
| cluster | character | fusion association |
|---|---|---|
| IC1 | IRX-type, differentiated | mixed |
| IC2 | IRX-type, undifferentiated | mixed |
| IC3 | HOXA-type | KMT2A::MLLT1 |
| IC4 | HOXA-type | KMT2A::MLLT3 |
| IC5 | HOXA-type | KMT2A::AFF1 |
Two findings from this deserve to sit at the front of any KMT2A::AFF1 dossier. First, the most refractory subgroup was IC2, the undifferentiated IRX-type cluster, which carried RAS pathway mutations in 100% of cases and the heaviest mutational burden of them; the number of RAS pathway mutations predicted prognosis 1. Second, KMT2A::AFF1 sat in its own HOXA-type cluster, distinct from MLLT1 and MLLT3 despite all three being HOXA-type. A one-size therapeutic model across KMT2A-r would collapse exactly the distinctions this analysis resolves.
Why 'undifferentiated equals worse' is not circular
It would be easy to dismiss 'the least differentiated cluster does worst' as a restatement of a known correlation. It is not, because in this cohort the undifferentiated cluster was also the one in which every case carried RAS pathway mutations 1. That converts a descriptive staging observation into a mechanistic and actionable one: the least differentiated state co-segregates with a specific, druggable signalling addiction. A MEK-plus-ATR combination has independently shown synergy in an MLL-AF4-plus-activated-RAS mouse and patient-derived xenograft model 4.
Definition and position in the hierarchy
Cell Ontology gives CL:0000837 (haematopoietic multipotent progenitor cell) and the marker-explicit CL:0002043 (CD34-positive, CD38-negative multipotent progenitor cell) 2. In human practice the MPP is separated from the HSC within the CD34-positive CD38-negative gate by loss of CD90/Thy-1 while remaining CD45RA-negative; acquisition of CD45RA moves the cell into the multilymphoid/LMPP compartment 3. So the MPP is the last stage before lineage priming, and the first stage that has surrendered long-term self-renewal.
That is precisely why a KMT2A fusion arising at or near this stage is dangerous: the cell already has full lineage breadth, and the fusion supplies back the self-renewal the cell has just given up.
The MPP compartment is itself perturbed by the fusion
This is not only a passive coordinate system. Ageing Mll-AF4-positive mice lost their B-lymphoid bias, showed an increased proportion of T-lymphoid cells in the spleen, and had decreased frequencies of haematopoietic stem and multipotent progenitor cells 5. A pre-leukaemic fusion therefore distorts the stem and MPP compartments long before overt leukaemia, which matters both for interpreting stage assignments in patients and for the idea that a pre-leukaemic clone might be detectable.
| ontology | CL:0000837 |
|---|---|
| location | Fetal liver and fetal bone marrow before birth; bone marrow after birth |
ontology alternatives
[ "CL:0002043" ]
markers
[ "Lineage-", "CD34+", "CD38-", "CD45RA-", "CD90/THY1-" ]
lineage potential
[ "all haematopoietic lineages, with limited self-renewal" ]
atlas refs
[
{
"label": "Infant KMT2A-r leukaemia multi-omics, n=84, five integrative clusters with single-cell differentiation inference",
"pmid": "36042201",
"title": "Multi-omics analysis defines highly refractory RAS burdened immature subgroup of infant acute lymphoblastic leukemia.",
"authors": "Isobe T, Takagi M, Sato-Otsubo A, et al.",
"venue": "Nat Commun",
"year": 2022,
"title_provenance": "Title/authors/venue backfilled from PubMed esummary on 2026-09-20 because this citation stored a bare PMID with no title, which made it unscoreable by verify-citations.py. BACKFILLING IS NOT VERIFICATION: it confirms the PMID resolves to a real paper, not that this is the right paper for the claim. That still requires reading."
},
{
"label": "Clonal human progenitor hierarchy map",
"pmid": "20543838",
"title": "Revised map of the human progenitor hierarchy shows the origin of macrophages and dendritic cells in early lymphoid development.",
"authors": "Doulatov S, Notta F, Eppert K, et al.",
"venue": "Nat Immunol",
"year": 2010,
"title_provenance": "Title/authors/venue backfilled from PubMed esummary on 2026-09-20 because this citation stored a bare PMID with no title, which made it unscoreable by verify-citations.py. BACKFILLING IS NOT VERIFICATION: it confirms the PMID resolves to a real paper, not that this is the right paper for the claim. That still requires reading."
}
]kmt2a clusters
[
{
"cluster": "IC1",
"label": "IRX-type differentiated"
},
{
"cluster": "IC2",
"label": "IRX-type undifferentiated",
"note": "most refractory; 100% RAS pathway mutated; heaviest RAS burden"
},
{
"cluster": "IC3",
"label": "HOXA-type MLLT1"
},
{
"cluster": "IC4",
"label": "HOXA-type MLLT3"
},
{
"cluster": "IC5",
"label": "HOXA-type AFF1"
}
]- The most refractory KMT2A-r infant subgroup is defined by RAS pathway mutation burden, a signalling addiction entirely separate from the menin-KMT2A interface, so menin inhibition may fail precisely where the need is greatest.
- Differentiation stage inferred by projection can be shifted by treatment, so a stage-based stratification measured at diagnosis may not describe the cell that relapses.
- A fusion arising near the MPP stage already has full lineage breadth, so any lineage-restricted immunotherapy is attacking one branch of a cell that has several.
- Are the IRX-type undifferentiated cases a distinct cell of origin, or the same origin with a RAS-driven de-differentiation? Single-cell genotype-plus-transcriptome on IC2 cases would distinguish these.
- Does menin dependency vary across the five integrative clusters? If IC2 is the most refractory and also the least differentiated, it is the group most likely to escape a menin inhibitor.
- Is the five-cluster solution reproducible in a non-Japanese cohort such as Interfant-06, and do the cluster-specific master transcription factors survive replication?
- Could MEK plus ATR inhibition be positioned specifically for the IC2 group rather than tested across unselected KMT2A-r ALL?
- E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
Multi-omics profiling of 84 infants with KMT2A-rearranged leukaemia, with single-sample and single-cell inference of haematopoietic differentiation, defined five integrative clusters corresponding to stages of B-lymphopoietic and early hemato-endothelial development (IRX-type differentiated, IRX-type undifferentiated, HOXA-type MLLT1, HOXA-type MLLT3, HOXA-type AFF1); the number of RAS pathway mutations predicted prognosis and the most refractory cluster, the undifferentiated IRX-type IC2, had 100% frequency and the heaviest burden of RAS pathway mutations.
Multi-omics analysis defines highly refractory RAS burdened immature subgroup of infant acute lymphoblastic leukemia · Isobe T, Takagi M, Sato-Otsubo A, Nishimura A, Nagae G, Yamagishi C, Tamura M, Tanaka Y, Asada S, Takeda R, Tsuchiya A, Wang X, Yoshida K, Nannya Y, Ueno H, Akazawa R, Kato I, Mikami T, Watanabe K, Sekiguchi M, Seki M, Kimura S, Hiwatari M, Kato M, Fukuda S, Tatsuno K, Tsutsumi S, Kanai A, Inaba T, Shiozawa Y, Shiraishi Y, Chiba K, Tanaka H, Kotecha RS, Cruickshank MN, Ishikawa F, Morio T, Eguchi M, Deguchi T, Kiyokawa N, Arakawa Y, Koh K, Aoki Y, Ishihara T, Tomizawa D, Miyamura T, Ishii E, Mizutani S, Wilson NK, Gottgens B, Miyano S, Kitamura T, Goyama S, Yokoyama A, Aburatani H, Ogawa S, Takita J · Nature Communications 2022What this evidence cannot show
Differentiation stage is inferred computationally by projecting bulk and single-cell profiles onto reference haematopoietic maps, so the stage assignment is a statistical nearest-neighbour rather than a measured cell identity. Cohort is predominantly Japanese, which bears on generalisability of subgroup frequencies. Cluster boundaries in a 5-cluster solution over 84 samples are not guaranteed to be stable in an independent cohort.
Evidence that argues against this (1)
- Inferred differentiation stage is not cell of origin. Sequencing shared mutations across an infant KMT2A-r B-ALL and its AML relapse in the same child placed the rearrangement before haematopoietic specification, and the authors state explicitly that 'cell of origin cannot be inferred from the transcriptional state'. An IC that looks like MPP identifies the state the blasts occupy, not the cell in which the translocation happened. Read: abstract only. Khabirova E et al. 2022
- E2established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database
Cell Ontology provides CL:0000837 (haematopoietic multipotent progenitor cell) and CL:0002043 (CD34-positive, CD38-negative multipotent progenitor cell).
Cell Ontology term lookup via the EMBL-EBI Ontology Lookup Service OLS4 API · EMBL-EBI Ontology Lookup Service; Cell Ontology consortium · Database 2026What this evidence cannot show
Ontology marker definitions are partly mouse-derived; CL treats a continuum as discrete classes.
Evidence that argues against this (1)
- The ontology terms describe a discrete stage that single-cell data do not resolve. Human HSCs acquire lineage bias gradually without traversing discrete hierarchically organised progenitor populations; unilineage cells emerge directly from a continuum of low-primed HSPCs. CL:0000837 and CL:0002043 are useful labels for gates, not evidence of a distinct cell. Read: abstract only. Velten L et al. 2017
- E3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
Within the human CD34-positive CD38-negative stem-cell compartment, acquisition of CD45RA with low or absent Thy-1/CD90 marks the multilymphoid progenitor, placing the CD45RA-negative CD90-negative fraction as the multipotent progenitor stage between HSC and lineage-primed progenitors.
Revised map of the human progenitor hierarchy shows the origin of macrophages and dendritic cells in early lymphoid development · Doulatov S, Notta F, Eppert K, Nguyen LT, Ohashi PS, Dick JE · Nature Immunology 2010What this evidence cannot show
Surface-marker gates impose discrete boundaries on a transcriptional continuum; later index-sorting and single-cell work has shown substantial functional heterogeneity within each of these gates.
Evidence that argues against this (2)
- CD45RA-negative CD90-negative CD34+CD38- is an impure and ontogeny-dependent gate. Fetal liver contains many oligopotent progenitors with intermingled myeloid, erythroid and megakaryocytic fates while adult marrow contains almost none, so the same markers do not isolate the same population across development. Read: abstract only. Notta F et al. 2016
- Independent work had to refine the HSC/MPP sorting strategy in human fetal liver and bone marrow to reach roughly 90% enrichment, which implies that the conventional marker gate is substantially contaminated; the same study found extensive epigenetic but not transcriptional priming in HSC/MPP, so a population defined by surface markers and scored by transcriptome can be chromatin-heterogeneous while looking uniform. Read: abstract only. Ranzoni AM et al. 2021
- E4moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper
Activating mutant RAS cooperates with MLL-AF4 to generate a highly aggressive serially transplantable B-ALL in mice, and dual inhibition of MEK and ATR is synergistic in both mouse-derived and patient-derived xenografts carrying both lesions.
Inhibition of MEK and ATR is effective in a B-cell acute lymphoblastic leukemia model driven by MLL-AF4 and activated RAS · Chu SH, Song EJ, Chabon JR, Minehart J, Matovina CN, Makofske JL, Frank ES, Ross K, Koche RP, Feng Z, Xu H, Krivtsov A, Nussenzweig A, Armstrong SA · Blood Advances 2018What this evidence cannot show
Preclinical only, with no clinical trial of this combination in infant KMT2A-r ALL reported. The mouse model uses retroviral expression of mutant RAS on an MLL-AF4 background, which is a stronger and more uniform RAS signal than the subclonal RAS mutations seen in patients.
Evidence that argues against this (1)
- Activating RAS cooperating with MLL-AF4 in MICE contradicts the human cord-blood result recorded elsewhere in this beat, where KRAS-G12V does not cooperate to immortalise or initiate. The species and the cell differ; the entry should say which it believes and why. The MEK/ATR synergy is also preclinical and RAS-mutant-selected. Oncofusion knowledge graph (internal analysis) 2026
- E5moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper
Ageing Mll-AF4-positive mice lose their B-lymphoid bias, show an increased proportion of splenic T-lymphoid cells, and have decreased frequencies of haematopoietic stem and multipotent progenitor cells, showing that the pre-leukaemic fusion perturbs the stem and MPP compartments before overt leukaemia.
Fetal liver Mll-AF4+ hematopoietic stem and progenitor cells respond directly to poly(I:C), but not to a single maternal immune activation · Malouf C, Ottersbach K · Experimental Hematology 2019What this evidence cannot show
Mouse pre-leukaemia model with a knock-in Mll-AF4 that does not faithfully reproduce human t(4;11) infant ALL; compartment frequency changes in ageing mice are also confounded by normal age-related haematopoietic drift.
Evidence that argues against this (1)
- A pre-leukaemic perturbation of the stem and MPP compartments in mice may not indicate where the human disease starts. In an independent Mll-AF4 knock-in the dominant outcome after long latency was B-cell lymphoma, not the human disease, and the authors concluded the fusion's lineage deregulation alone is insufficient for malignancy. Perturbation of a compartment and transformation of a compartment are different claims. Read: abstract only. Chen W et al. 2006
- part of →Common myeloid progenitor (CMP) — the fork at which KMT2A leukaemia almost always turns one way
CMP is the myeloid-restricted stage downstream of the MPP
gives rise to (next stage in the hierarchy)
gives rise to; CD45RA acquisition marks the lymphoid-primed stage
- related to →The differentiation block — what it actually is, operationally, in KMT2A::AFF1 leukaemia
the undifferentiated IRX-type infant cluster maps near this stage and is the most refractory
reproduces the five-cluster differentiation-stage classification and its RAS burden association
- related to →HOXA-high versus HOXA-low KMT2A-rearranged leukemia — the split that breaks the one-mechanism story
the five infant integrative clusters split into HOXA-type and IRX-type with different differentiation stages
in multipotent progenitors Mef2c is required for proper expression of key lymphoid regulators, downstream of PU.1 (murine; PMID 19169261)