Oncofusion

start / graph / gene-samhd1

Proteinstrong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.Q49-metabolism

SAMHD1 - the dNTP hydrolase that is the real biomarker behind the IMPDH story, and the reason the KMT2A framing of that story does not hold

also called SAMHD1 · MOP5 · SAM domain and HD domain-containing protein 1 · dNTPase · Q9Y3Z3

SAMHD1 is an enzyme that chews up the building blocks of DNA, including the chemotherapy drug cytarabine after the cell has activated it - so cells with a lot of SAMHD1 resist that drug. Two 2026 papers show that blocking a metabolic enzyme called IMPDH starves SAMHD1 of the signal that switches it on, which makes cytarabine work again; the same papers found this has nothing to do with whether the leukaemia carries a KMT2A fusion.

The same thing, for a scientist

SAMHD1 (UniProt Q9Y3Z3, 626 aa, primary gene name read back 2026-09-20) is a deoxynucleoside triphosphate triphosphohydrolase whose activity is allosterically gated by GTP and dGTP. It hydrolyses the active triphosphate metabolites of nucleoside-analogue chemotherapies, including ara-CTP, and so constitutes a pharmacological barrier to cytarabine and fludarabine. Two independent 2026 studies converge: inhibiting IMPDH depletes GTP/dGTP, removes the allosteric activation of SAMHD1, raises active triphosphate metabolites and sensitises leukaemic cells to ara-C - **in SAMHD1-proficient but not SAMHD1-deficient cells** [E1][E2]. SAMHD1 belongs in this beat because it is the node that converts a *metabolic* intervention (guanine-nucleotide depletion) into a *chemosensitisation* phenotype, and because it is the measured effect modifier where KMT2A status is not: Klootsema et al. tested the KMT2A link directly and report that sensitivity to IMPDH inhibitors was **independent of KMT2A status**, with the synergy absent in ALL cell lines [E1]. This entry exists partly to keep the graph honest about a paper whose title names KMT2A-rearranged leukaemia and whose finding is that the fusion does not matter.

What it looks like
loading AF-Q9Y3Z3-F1…
AlphaFold AF-Q9Y3Z3-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
Why a dNTPase is a metabolism entry

SAMHD1 is usually filed under innate immunity or nucleotide catabolism, not cancer metabolism. It sits here because the only route by which the guanine-nucleotide literature reaches the clinic in this disease runs through it. The chain is:

IMPDH inhibitor -> GTP and dGTP fall -> SAMHD1 loses its allosteric activator -> ara-CTP is no longer hydrolysed -> cytarabine works better 12.

Every step is metabolic except the last. And the biomarker this predicts is SAMHD1 protein, not a fusion, not a transcriptional signature. That is an unusually well specified, prospectively testable prediction for a repurposing proposal, and it is the one this graph should carry rather than the KMT2A framing.

The correction this entry makes to a plausible reading

The Cell Cycle paper is titled Targeting IMPDH to inhibit SAMHD1 in KMT2A-rearranged leukaemia 1. A title-only reading gives: IMPDH inhibitors are a KMT2A-rearranged strategy. The abstract gives the opposite: "We found that sensitivity to IMPDHi was independent of KMT2A status", and "IMPDHi synergized with ara-C and F-ara-A in a SAMHD1-dependent manner in a subset of AML cells, but not in acute lymphoblastic leukaemia cell lines" 1.

This is the same failure shape this project logs elsewhere as a title-restatement error, and it is worth naming because the earlier, celebrated IMPDH paper in this disease is genuinely MLL-fusion-framed [see gene-impdh2 E2], so a reader moving from one to the other would reasonably assume the second confirms the first. It does not. It measures the thing the first did not measure, and finds no genotype effect.

Bound on that negative: it is a cell-line and primary-sample pharmacology result, the abstract gives no n for the KMT2A-rearranged versus wild-type arms, and no infant B-lineage primary material or PDX was tested. It is an absence of a detected difference, not a demonstrated equivalence.

Regulated nuclear import, and a damage-conditional chromatin address (Q87, 2026-09-21)

Subcellular. Predominantly nuclear by ACTIVE import - KPNA2 plus a classical N-terminal NLS (14KRPR17 in PMID 24712655; the same four residues are numbered 11KRPR14 elsewhere) - with a cytoplasmic pool when the NLS or karyopherin availability is disrupted. HPA v25.1 IF: nucleoplasm main, plasma membrane additional (unexplained).

Subnuclear: sites of DNA double-strand breaks. SAMHD1 is recruited to DSBs and recruits CtIP there to promote end resection and HR, dNTPase-independently (PMID 28834754). Damage-conditional chromatin, not a standing compartment; no nucleolus, PML body or speckle.

Cell type and state. Myeloid cells and quiescent CD4 T cells - genuinely haematopoietic, and about NON-CYCLING cells. Second state: the DNA-damage response.

The compartment does not carry the function. Cytoplasmically relocalised SAMHD1 retains antiviral dNTPase activity, so 'nuclear' cannot be used to attribute a phenotype to the metabolic arm.

Shown for fusion. No. The single fusion-scoped hit was retrieved and rejected - a myeloma sequencing survey matching on 'MLL' in a mutation list.

Facts and figures
uniprotQ9Y3Z3
length AA626
alphafold IDAF-Q9Y3Z3-F1
pathwaydNTP catabolism; allosterically activated by GTP/dGTP
resolution notegene_exact:SAMHD1 AND organism_id:9606 AND reviewed:true returned 1 reviewed row on 2026-09-20; gene_primary read back as SAMHD1; length 626 aa asserted.
role in beateffect modifier converting guanine-nucleotide depletion into nucleoside-analogue chemosensitisation
kmt2a relationshipexplicitly measured as independent of KMT2A status (PMID 41399259)
evidence lineageAML cell lines and primary AML; ALL cell lines negative
subnuclear localizationSITES OF DNA DOUBLE-STRAND BREAKS. SAMHD1 is recruited to DSBs and recruits CtIP there via a conserved C-terminal domain to facilitate end resection and homologous recombination, independently of its dNTPase activity (PMID 28834754; UniProt Q9Y3Z3 'Chromosome', note 'Localizes to sites of DNA double-strand breaks in response to DNA damage'). This is a damage-CONDITIONAL chromatin address, not a standing compartment, and no nucleolar, PML-body or speckle pool has been reported. Nuclear residence itself is actively maintained through KPNA2 and an N-terminal NLS (PMID 24712655), so the nuclear/cytoplasmic split is regulated rather than fixed.
cell stateQUIESCENCE / NON-DIVISION is the state SAMHD1 is defined against, and it cuts the opposite way from a proliferation dependency: SAMHD1's dNTP-depleting activity is what makes resting myeloid cells and quiescent CD4 T cells refractory to reverse transcription, which is a property of cells that are NOT replicating DNA. SAMHD1[tiab] AND (quiescent[tiab] OR nondividing[tiab] OR "non-dividing"[tiab] OR differentiated[tiab]) returns 93 records (2026-09-21, quoted as a literature size, not title-checked). The second state is DNA-DAMAGE RESPONSE, where SAMHD1 acts at breaks.
shown for fusionNO - nobody has looked in any fusion. Measured 2026-09-21: SAMHD1[tiab] AND the eight-term fusion clause AND the ten-term localisation clause returns 1 record, PMID 35770320, which was retrieved and read: it is a next-generation-sequencing survey of plasma cell disorders in a Thai population, matches on 'MLL' appearing in a mutation list, and contains no localisation experiment and no KMT2A fusion result. Screened count on topic: 0.
depmap kmt2a selectivity
{
  "chronos_mean_kmt2a_rearranged": -0.2396,
  "chronos_mean_other_leukaemia": -0.2553,
  "difference": 0.0157,
  "pan_essentiality_all_lines": -0.2012,
  "selectivity_rank": 11199,
  "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."
}
localization
{
  "subcellular": "PREDOMINANTLY NUCLEAR, by an ACTIVE and mapped import route - not by default. SAMHD1 preferentially uses karyopherin alpha 2 (KPNA2) and a classical N-terminal nuclear localization signal, given in that paper as 14KRPR17, to enter the nucleus; reducing karyopherin availability or disrupting the NLS moves the protein to the cytoplasm, and nuclear residence is what confers sensitivity to VpxMAC-induced ubiquitination and proteasomal degradation (PMID 24712655). UniProt Q9Y3Z3 lists Nucleus on six coded experimental citations and Chromosome on one. HPA v25.1 IF, reliability Supported, gives Nucleoplasm main with PLASMA MEMBRANE additional - a pool neither UniProt nor the import literature records, and which should be treated as unexplained rather than as established.",
  "subnuclear": "SITES OF DNA DOUBLE-STRAND BREAKS - a damage-induced chromatin address rather than a standing compartment. SAMHD1 is recruited to DSBs, complexes with CtIP through a conserved C-terminal domain and recruits CtIP to breaks to facilitate end resection and homologous recombination, and this is a dNTPase-INDEPENDENT function: a cancer-associated mutant with impaired CtIP interaction fails to rescue the end-resection defect while a dNTPase-dead mutant does (PMID 28834754). UniProt records the corresponding 'Chromosome' term with the note 'Localizes to sites of DNA double-strand breaks in response to DNA damage'. No nucleolus, no PML body, no speckle: the subnuclear answer here is damage-conditional chromatin.",
  "the_partition_is_regulated_not_fixed": "A single 'nucleus' answer would be wrong twice over. The import is an active, saturable, karyopherin-dependent process that can be broken; and the functional literature the protein is famous for turns on the cytoplasmic consequence of nuclear residence - SAMHD1's restriction of lentiviral reverse transcription happens in the CYTOPLASM while the protein itself is nuclear, and NLS-deleted SAMHD1 relocalised to the cytoplasm retains and in some species enhances antiviral function.",
  "how_shown": "GST pull-down and RNA-silencing of karyopherins with NLS mutagenesis and imaging (PMID 24712655); recruitment to DSBs with CtIP co-complex mapping and separation-of-function mutants (PMID 28834754); HPA antibody IF",
  "node": "localization-chromatin-bound-versus-soluble-fraction-kmt2a"
}
cell type
{
  "haematopoietic_answer": "MYELOID CELLS AND QUIESCENT CD4 T CELLS. SAMHD1 is described as a nuclear protein that inhibits HIV-1 infection in myeloid cells and in quiescent CD4 T cells by lowering the intracellular dNTP pool below what reverse transcription needs (PMID 24712655) - a cell-type answer that is genuinely haematopoietic, and that is about NON-CYCLING cells, which is the opposite of the proliferative context this graph usually reasons in.",
  "expression_atlas": "HPA v25.1: RNA tissue specificity Low; IF reliability Supported (nucleoplasm main, plasma membrane additional)."
}
How this could bite us back
  • Making cytarabine work better by disabling a dNTP hydrolase makes it work better everywhere SAMHD1 is expressed, including in normal marrow and gut. The proposed mechanism has no tumour-selective step in it.
  • The intervention that reaches SAMHD1 here is an approved immunosuppressant given to a myelosuppressed child alongside cytarabine; the infection risk is additive with the very chemotherapy it is meant to potentiate.
  • If SAMHD1 is the biomarker, a patient can lose the benefit by losing SAMHD1 - and SAMHD1-low disease is exactly the disease already sensitive to cytarabine, so the strategy is structurally confined to the resistant subgroup and cannot be assessed by an unselected trial.
Still open
  1. What is SAMHD1 protein abundance in infant KMT2A::AFF1 blasts at diagnosis? If it is low, the entire IMPDH-plus-cytarabine strategy is inapplicable to the index disease for a reason that is measurable today.
  2. Is the ALL negative a SAMHD1-expression negative or a genuine lineage negative? The two studies did not report SAMHD1 levels in the ALL lines that failed to synergise, so the simplest explanation has not been excluded.
  3. Does SAMHD1 status modify the LEDGF/menin arm of the guanine-nucleotide mechanism, or only the nucleoside-analogue arm? Nobody has tested a menin inhibitor plus an IMPDH inhibitor stratified by SAMHD1.
Every source behind this entry (5)
  1. E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    Sensitivity to IMPDH inhibitors was independent of KMT2A status in AML cell lines and primary AML samples; IMPDH inhibitors synergised with cytarabine and fludarabine in a SAMHD1-dependent manner in a subset of AML cells but not in acute lymphoblastic leukaemia cell lines; mechanistically IMPDH inhibition depleted the allosteric SAMHD1 activators GTP and dGTP, thereby increasing active triphosphate metabolites in SAMHD1-proficient but not SAMHD1-deficient cells.

    Targeting IMPDH to inhibit SAMHD1 in KMT2A-rearranged leukaemia. · Klootsema Y, Tsesmetzis N, Sharma S, ..., Herold N · Cell Cycle 2026
    PMID 41399259DOI
    What this evidence cannot show

    Abstract read this session; full text not retrieved. The KMT2A-independence claim is an absence of a detected difference in a panel of unstated size, not a demonstrated equivalence, and the ALL negative is in cell lines rather than infant KMT2A::AFF1 primary material or PDX. The paper describes mycophenolic acid as a hit from a prior phenotypic screen for indirect SAMHD1 inhibitors, so the screening context preselected for SAMHD1-linked activity.

    Evidence that argues against this (1)
    • The earlier MLL-fusion IMPDH study reports in vitro susceptibility of MLL-fusion AMLs and in vivo suppression of MLL-AF9-driven AML development, endpoints this study does not test; an absence of genotype-linked differential sensitivity in a cell-line viability assay does not exclude a genotype-linked effect on leukaemia initiation. Liu X et al. 2023
  2. E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    SAMHD1 hydrolyses ara-CTP back to its inactive form, constituting a barrier to cytarabine efficacy; a screen of drugs targeting nucleotide biosynthetic enzymes identified IMPDH inhibition as sensitising AML cell lines to ara-C in a SAMHD1-dependent manner, and the approved IMPDH inhibitors mycophenolic acid and ribavirin imbalance dNTP pools and increase ara-C efficacy in SAMHD1-proficient but not SAMHD1-deficient leukaemic cells.

    PMID 41711315DOI
    What this evidence cannot show

    Abstract read this session. AML cell lines; the abstract makes no KMT2A claim, so this is evidence about the SAMHD1-ara-C axis and not about the index disease. Ribavirin has activities beyond IMPDH, so the two-drug concordance is weaker than it looks. The result is a chemosensitisation, not a single-agent effect.

    Evidence that argues against this (1)
    • The companion study reached the same SAMHD1-dependence conclusion while explicitly finding no synergy in acute lymphoblastic leukaemia cell lines, which bounds the strategy away from the lymphoid index disease of this graph even where SAMHD1 is expressed. Klootsema Y et al. 2026
  3. E3established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database

    UniProt reviewed human record Q9Y3Z3 (SAMH1_HUMAN) has primary gene name SAMHD1, gene synonym MOP5, length 626 aa, and recommended protein name 'Deoxynucleoside triphosphate triphosphohydrolase SAMHD1 (dNTPase)'.

    What this evidence cannot show

    Resolved 2026-09-20 with gene_exact:SAMHD1 AND organism_id:9606 AND reviewed:true; exactly one reviewed row and gene_primary read back as SAMHD1. Unlike the IMPDH pair, SAMHD1 has no close paralogue in this length range, so the accession-plus-length assertion is a real defence here rather than a nominal one. The record fixes sequence and symbol on the date read and carries no information about expression in KMT2A-rearranged blasts.

    Evidence that argues against this (1)
    • A database record cannot establish the quantity that actually matters clinically here, which is SAMHD1 protein level in an individual patient's blasts at diagnosis. The two studies cited make SAMHD1 proficiency the effect modifier, so the entry's translational claim depends on a measurement nobody has yet made prospectively in KMT2A-rearranged paediatric disease. Yague-Capilla M et al. 2026
  4. E4strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    SAMHD1's nuclear residence is actively maintained rather than default: it preferentially uses karyopherin alpha 2 (KPNA2) and a classical N-terminal nuclear localization signal, given as 14KRPR17, to enter the nucleus by the classical karyopherin alpha/beta1 pathway, and that nuclear localisation is what confers sensitivity to VpxMAC-induced ubiquitination and proteasomal degradation. SAMHD1 is described in the same work as the nuclear protein that inhibits HIV-1 in myeloid cells and quiescent CD4 T cells.

    PMID 24712655DOI
    What this evidence cannot show

    The system is HIV restriction in myeloid and T-cell models, not leukaemia, and the NLS mapping uses mutant constructs. Note that the residue numbering of the NLS is not consistent across the literature - a comparative study of feline and bovine SAMHD1 gives the human motif as 11KRPR14 - so the motif is the same four residues and the index differs by three; the string quoted here is the one this paper gives. HPA additionally reports a plasma-membrane pool that no mechanistic study explains.

    Evidence that argues against this (1)
    • The natural reading - SAMHD1 is nuclear, therefore its activity is nuclear - is contradicted by the protein's best-characterised function. Its restriction of lentiviral reverse transcription operates on the CYTOPLASMIC dNTP pool, and comparative work shows that NLS-deleted SAMHD1 relocalised to the cytoplasm RETAINS antiviral function and in the feline protein restricts Vpx-encoding SIV and HIV-2 MORE efficiently than the wild-type nuclear form. So nuclear localisation is not required for the enzymatic job; what it buys is something else - accessibility to Vpx-mediated degradation, and, separately, the DNA-end-resection role. An entry that records 'nucleus' and stops would mispredict every experiment in which the compartment is manipulated. Wang C et al. 2021
  5. E5strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    SAMHD1's subnuclear address is damage-conditional chromatin: it is recruited to DNA double-strand breaks, complexes with CtIP through a conserved C-terminal domain and recruits CtIP to breaks to facilitate end resection and repair by homologous recombination. The function is dNTPase-INDEPENDENT - a cancer-associated mutant with impaired CtIP interaction fails to rescue the end-resection defect of SAMHD1 depletion, while a dNTPase-inactive mutant does - and SAMHD1 deficiency or Vpx-mediated degradation causes hypersensitivity to DSB-inducing agents.

    SAMHD1 Promotes DNA End Resection to Facilitate DNA Repair by Homologous Recombination. · Daddacha W, Koyen AE, Bastien AJ, Head PE, Dhere VR, Nabeta GN, et al. · Cell Rep 2017
    PMID 28834754DOI
    What this evidence cannot show

    Recruitment to breaks is measured in non-haematopoietic human cell systems using induced-damage assays, so the address exists only in the presence of damage and says nothing about where SAMHD1 sits in an unperturbed leukaemic blast. This is an occupancy/recruitment result rather than a named subnuclear compartment, and it does not identify a body, domain or territory.

    Evidence that argues against this (1)
    • This result competes with, rather than complements, the reason this entry exists: the beat's interest in SAMHD1 is as the dNTP hydrolase behind the IMPDH story, and the DSB role is explicitly dNTPase-independent. That means the two activities can be separated by mutation, and a SAMHD1 phenotype in a leukaemia screen could belong to either - the knockout removes both. Worse for the metabolic reading, the protein's nuclear residence is not needed for the dNTPase job at all, since cytoplasmically relocalised SAMHD1 still restricts reverse transcription; so a compartment-based argument cannot be used to attribute a phenotype to the dNTPase arm. Wang C et al. 2021
Connected to