Granulocyte-monocyte progenitor (GMP) — the committed cell an MLL fusion can turn back into a stem cell
also called GMP · granulocyte-macrophage progenitor · granulocyte monocyte progenitor cell · Lin-CD34+CD38+IL-3Ra+CD45RA+
The granulocyte-monocyte progenitor is a bone-marrow cell that has already decided to become one of two kinds of infection-fighting white cell and has given up the ability to renew itself indefinitely. The landmark finding in this disease is that an MLL fusion can put that ability back: a committed, short-lived cell becomes a self-renewing leukaemia stem cell without otherwise changing what it is.
The same thing, for a scientist
The GMP (CL:0000557) is the myeloid-committed progenitor downstream of the common myeloid progenitor, prospectively isolable in human marrow and cord blood as Lin-CD34+CD38+IL-3R-alpha-positive CD45RA-positive, producing granulocytes and monocytes but not megakaryocytes or erythrocytes [E1]. It is the single most consequential normal cell type in the KMT2A literature because of one experiment: leukaemia stem cells generated by introducing MLL-AF9 into committed GMP transferred leukaemia at **four cells** per secondary recipient while retaining the immunophenotype and global expression profile of normal GMP, with only a limited HSC-associated self-renewal programme re-activated [E2]. That result established that a cancer stem cell need not resemble a tissue stem cell, and it is the empirical origin of this graph's `cell-state-stem-like-self-renewal`. The GMP is also the destination state in therapy-associated lineage switch and the state most sensitive to menin inhibition ex vivo, which is the subject of `cell-state-gmp-like`.
Where the GMP sits and what it would be doing if it were healthy
Blood cells are made in a branching sequence. A stem cell makes a multipotent progenitor; that makes a common myeloid progenitor (CMP) or a lymphoid-primed progenitor; the CMP makes either a megakaryocyte-erythroid progenitor (MEP) or a granulocyte-monocyte progenitor (GMP). The GMP's job is to produce the two front-line phagocytes: neutrophils (and the other granulocytes) and monocytes. Beyond that point the cell is committed; it will divide a limited number of times, produce mature progeny, and its lineage will end.
In humans all three myeloid progenitors sit in the lineage-negative CD34+CD38+ fraction of adult bone marrow and cord blood, and they are separated by two markers: the interleukin-3 receptor alpha chain and CD45RA. The GMP is the IL-3R-alpha-positive CD45RA-positive fraction; the MEP is IL-3R-alpha-negative CD45RA-negative; the CMP is IL-3R-alpha-low CD45RA-negative 1.
So the GMP is defined by two things at once: it can make exactly two mature cell types, and it cannot renew itself. Both matter for what a KMT2A fusion does to it.
The experiment that made the GMP central to this disease
Before 2006 the dominant assumption was that a cancer stem cell had to be, or had to resemble, a normal tissue stem cell. Introducing MLL-AF9 into committed granulocyte-macrophage progenitors overturned that. The resulting leukaemia stem cells:
- transferred leukaemia to secondary recipients when only four cells were transplanted;
- possessed an immunophenotype and global gene expression profile very similar to that of normal GMP;
- had re-activated a subset of genes highly expressed in normal HSCs, but had not undergone widespread reprogramming 2.
The conclusion the authors drew is the one this graph builds on: LSCs can be generated from committed progenitors without wholesale reprogramming, and a self-renewal programme expressed in an abnormal context may be targetable. That last clause is the intellectual ancestor of the menin-interface strategy — the drug is not trying to kill a stem cell, it is trying to switch off a self-renewal module running in a cell that should not have one.
A practical corollary that is easy to miss: if the LSC looks like a normal GMP, then surface-marker-based purging or CAR targeting of the LSC will also hit normal GMP, and the normal GMP is the source of neutrophils and monocytes. Any GMP-directed immunotherapy in this disease is a myeloablative strategy by construction.
The GMP in KMT2A-rearranged disease today
Three separate strands now converge on this compartment.
Lineage switch lands here. Profiling oncoprotein-target sites across 36 MLL-rearranged samples, including three that lineage-switched on therapy, found that in switching samples the oncoprotein levels are reduced and the oncoprotein preferentially activates granulocyte-monocyte progenitor genes 3.
Menin inhibition works best here. A large-scale drug screen across primary AML cells, integrating genetics and quantitative proteomics with drug sensitivity, found GMP-like AML display the greatest sensitivity to the menin-KMT2A inhibitor bleximenib 4.
The two together make a paradox that this graph has already recorded as hyp-gmp-destination-paradox-menin: if therapy pushes cells toward a GMP-like state and the GMP-like state is where menin inhibition works best, a lineage switch on a menin inhibitor should predict response rather than failure. That hypothesis now has a cell type underneath it.
| ontology | CL:0000557 |
|---|---|
| location | Bone marrow after birth; fetal liver and fetal bone marrow before birth |
| immediate upstream | cell-type-cmp |
markers
[ "Lineage-", "CD34+", "CD38+", "IL3RA/CD123+", "CD45RA+" ]
lineage potential
[ "neutrophil and other granulocytes", "monocyte and macrophage" ]
immediate downstream
[ "cell-type-monoblast", "cell-type-monocyte" ]
kmt2a relevance
{
"transformation": "MLL-AF9 introduced into committed mouse GMP generates LSCs that transfer leukaemia at 4 cells while retaining the GMP expression profile",
"lineage_switch_destination": "oncoproteins in lineage-switching samples preferentially activate GMP genes",
"drug_sensitivity": "GMP-like AML shows the greatest ex vivo sensitivity to bleximenib"
}atlas refs
[
{
"label": "Prospective isolation of human CMP, GMP and MEP",
"pmid": "12193648",
"title": "Prospective isolation of human clonogenic common myeloid progenitors.",
"authors": "Manz MG, Miyamoto T, Akashi K, et al.",
"venue": "Proc Natl Acad Sci U S A",
"year": 2002,
"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": "MLL-AF9 transformation of committed GMP into LSC",
"pmid": "16862118",
"title": "Transformation from committed progenitor to leukaemia stem cell initiated by MLL-AF9.",
"authors": "Krivtsov AV, Twomey D, Feng Z, et al.",
"venue": "Nature",
"year": 2006,
"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."
}
]- If the leukaemia stem cell is immunophenotypically a normal GMP, any antibody, CAR or engager directed at it will also destroy the normal granulocyte and monocyte supply, and the resulting neutropenia is not a manageable off-target effect but the direct consequence of the mechanism.
- Menin inhibition is most effective in the GMP-like state, so the drug is strongest precisely where the leukaemia most resembles a normal committed progenitor, which is where on-target effects on normal myelopoiesis are hardest to avoid.
- A self-renewal module bolted onto a committed progenitor can in principle be bolted on again by a different route; removing the menin-KMT2A interface removes one route, not the capacity.
- Is the chromatin-defined GMP-like programme of lineage-switching samples the same thing as the proteomically defined GMP-like class that is bleximenib-sensitive? Scoring both on the same cells is the experiment, and until it is done the paradox in hyp-gmp-destination-paradox-menin may be an artefact of two different definitions.
- Does human GMP transformation by a KMT2A fusion reproduce the mouse result? The four-cell transfer experiment has never been repeated with human GMP and a physiological-level fusion.
- Because the LSC resembles a normal GMP, is there any surface antigen that separates them? If not, every GMP-directed immunotherapy in this disease is myeloablative by design.
- Does the GMP gate defined in adult marrow identify the corresponding cell in fetal liver, where infant KMT2A-rearranged leukaemia actually begins?
- E1established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper
Human clonogenic common myeloid progenitors and their downstream granulocyte/macrophage and megakaryocyte/erythrocyte progenitors all reside in the lineage-negative CD34+CD38+ fraction of adult bone marrow and cord blood and are prospectively separable by expression of the IL-3 receptor alpha chain and CD45RA, with the GMP being IL-3R-alpha-positive CD45RA-positive.
Prospective isolation of human clonogenic common myeloid progenitors · Manz MG, Miyamoto T, Akashi K, Weissman IL · Proceedings of the National Academy of Sciences USA 2002What this evidence cannot show
Prospective isolation by surface markers with in vitro clonogenic readout; purity is not absolute and the gates were defined in adult marrow and cord blood, not fetal tissue. The scheme imposes discrete compartments on what later single-cell work resolves as a continuum. Read: abstract only.
Evidence that argues against this (1)
- The discrete CMP-GMP-MEP hierarchy does not survive ontogeny-resolved single-cell analysis. Fetal liver contains many oligopotent progenitors with intermingled myeloid, erythroid and megakaryocytic fates while adult marrow contains almost none, so the same IL-3R-alpha/CD45RA gates do not isolate the same population across development. Since KMT2A-rearranged infant leukaemia arises prenatally, the adult-derived GMP gate may not describe the cell of origin at all. Read: abstract only. Notta F et al. 2016
- E2established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper
Leukaemia stem cells isolated from leukaemias initiated in committed granulocyte-macrophage progenitors by introduction of MLL-AF9 transferred leukaemia to secondary recipient mice when only four cells were transferred, and possessed an immunophenotype and global gene expression profile very similar to normal granulocyte-macrophage progenitors, with a subset of genes highly expressed in normal haematopoietic stem cells re-activated; LSCs can therefore be generated from committed progenitors without widespread reprogramming of gene expression.
Transformation from committed progenitor to leukaemia stem cell initiated by MLL-AF9 · Krivtsov AV, Twomey D, Feng Z, Stubbs MC, Wang Y, Faber J, et al., Armstrong SA · Nature 2006What this evidence cannot show
Retroviral MLL-AF9 in sorted mouse GMP followed by transplantation. The fusion is expressed from a retroviral promoter at non-physiological level, and mouse GMP are not human GMP. It establishes that a committed progenitor CAN be transformed, not that human KMT2A-rearranged leukaemia originates in a GMP.
Evidence that argues against this (1)
- Transcriptional or immunophenotypic resemblance to a GMP is not evidence of GMP origin. Sequencing shared mutations across an infant KMT2A-rearranged B-ALL and its AML relapse in the same child placed the rearrangement before haematopoietic specification, and that study states explicitly that cell of origin cannot be inferred from transcriptional state. A GMP-like LSC identifies the state the blasts occupy, not the cell in which the translocation occurred. Read: abstract only. Khabirova E et al. 2022
- E3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
Across 36 MLL-rearranged leukaemia samples profiled for oncoprotein-target sites, including three that underwent therapy-associated lymphoid-to-myeloid lineage switch, oncoprotein levels in the switching samples were reduced and the oncoproteins preferentially activated granulocyte-monocyte progenitor genes.
MLL oncoprotein levels influence leukemia lineage identities · Janssens DH, Duran M, Otto DJ, Wu W, Xu Y, Kirkey D, et al., Henikoff S · Nature Communications 2024What this evidence cannot show
Only three of the 36 samples lineage-switched, so the GMP destination rests on a very small subgroup; CUT&Tag occupancy is a chromatin readout, not a cell-identity assay, and no functional demonstration that the switched cells behave as GMP is offered. Read: abstract only.
Evidence that argues against this (1)
- The clinical series do not read lineage switch as movement into a favourable, drug-sensitive state. Lineage switch after CD19-directed therapy in KMT2A-rearranged leukaemia is consistently reported with poor outcomes, which is the opposite prognostic sign from what a move into the most menin-sensitive state would predict. The tension is recorded in this graph as hyp-gmp-destination-paradox-menin and is unresolved; the GMP destination finding should not be read as reassurance. Read: abstract only. Hogeling SM et al. 2025
- E4strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
A large-scale drug screen comparing genetic alterations and quantitative proteomics with drug sensitivity in primary AML patient cells revealed that granulocyte-macrophage progenitor (GMP)-like AML display the greatest sensitivity to the menin-KMT2A inhibitor JNJ-75276617 (bleximenib).
Bleximenib, the novel menin-KMT2A inhibitor JNJ-75276617, impairs long-term proliferation and immune evasion in acute myeloid leukemia · Hogeling SM, Le DM, La Rose N, Kwon MC, Wierenga ATJ, Van den Heuvel FAJ, et al., Schuringa JJ · Haematologica 2025What this evidence cannot show
Ex vivo sensitivity in primary AML cells, not clinical response; 'GMP-like' is a proteomic cluster assignment defined within this study and not obviously the same construct as the chromatin-defined GMP-like programme of E3. Sensitivity was also found in NPM1c/DNMT3A-mutant AML, so GMP-like is not a KMT2A-specific predictor. Read: abstract only.
Evidence that argues against this (1)
- If GMP-like AML is the most menin-sensitive class and NPM1c AML is also sensitive, then GMP-like status may be tracking the underlying HOXA/MEIS1 dependency common to both rather than a cell-identity property. Loss of PRC1.1 subunits confers menin-inhibitor resistance through reactivation of non-canonical menin targets including MYC, entirely independently of differentiation state, showing that menin sensitivity is set by chromatin context and not only by where the cell sits in the myeloid hierarchy. Read: abstract only. Zhou X et al. 2024
MLL-AF9 in a committed GMP re-activates a limited HSC self-renewal programme
the GMP is the immediate progenitor of the monocyte
supplies the cell type that hypothesis rested on without one
- related to →Lineage switch in KMT2A-rearranged leukemia — the same clone changes what kind of blood cell it is
lineage-switching samples preferentially activate GMP genes
GMP-like AML shows the greatest ex vivo sensitivity
the state is scored against this normal cell type, and is not the same thing
GMP is the immediate progenitor of the monoblast
- ← part ofCommon myeloid progenitor (CMP) — the fork at which KMT2A leukaemia almost always turns one way
GMP is one of the two branches downstream of the CMP gate
required in KMT2A-rearranged lines of both lineages (-0.559 myeloid, -0.683 B-lineage); no lineage restriction
myelo-erythroid progenitor compartment in this graph's vocabulary
the class is the authors' GMP-like gene export, scored as occupancy not expression
the same Pu.1-driven programme drives myelopoiesis and neutrophilia; the granulocyte-monocyte route is the destination of the shift
the LSC compartment in the murine MLL-AF9 model this result comes from is granulocyte-macrophage-progenitor-like; the derepressed targets are Cebpa and Cebpd, the myeloid differentiation transcription factors of exactly that compartment
REQUIRED IN: Smarcd2-deficient mice fail to generate mature neutrophils and eosinophils and SMARCD2-SWI/SNF is needed for CEBP-epsilon recruitment to secondary granule promoters (murine normal granulopoiesis; PMID 28369034)
RESTRAINS: relB-null mice show myeloid hyperplasia and splenomegaly from extramedullary haematopoiesis, so RelB limits the myeloid compartment in an animal with no fusion - the same direction as this entry's MLL-AF9 suppression result (murine; PMID 7834753)
FUT4 is the MYELOID alpha-1,3-fucosyltransferase, functional in human leukocytic HL-60 cells and murine bone-marrow neutrophils - the only genuinely lineage-restricted expression pattern in this tranche (PMID 23192350)
Rad21 haploinsufficiency accumulates granulocyte-macrophage progenitors in murine bone marrow
strong nuclear PBX2 protein in myeloid cells (PMID 10082572); contradicted at RNA level by HPA Blood Atlas 'not detected in immune cells'
PU.1 dose sets granulocyte-macrophage output; 20% of normal PU.1 in mouse gives AML and 80% loss gives a precancerous abnormal precursor pool still G-CSF-responsive with M-/GM-CSF pathways disrupted (PMID 15146183) (Q91 2026-09-22)
SBDS RNA and protein are expressed in human haematopoietic progenitors and in PLB-985, and expression FALLS during neutrophil differentiation; SDS patient progenitors show reduced differentiation and proliferation (PMID 19759903) (Q91 2026-09-22)