Oncofusion

start / graph / cell-state-replication-stress-icl-repair

Cell statemoderate confidence 3 of 5: One strong study, or several smaller consistent ones.Q73-vocabulary

Replication stress and interstrand-crosslink repair as a cell state - an ATR dependency measured in mouse MLL-rearranged AML, and the same stress implicated in breaking the MLL locus in the first place

also called replication stress · replicative stress state · stalled-fork state · interstrand-crosslink repair state · Fanconi/ICL-repair-dependent state · fork protection state

Copying DNA is not smooth: forks stall, and a repair crew has to restart them before the chromosome breaks. This entry records the evidence that leukaemia cells carrying an MLL/KMT2A rearrangement live under that kind of stress, and that the same stress is one of the ways the rearrangement is thought to be created.

The same thing, for a scientist

This node exists because five entries in this graph - `gene-fancd2`, `gene-fanci`, `gene-fancf`, `gene-brip1` and `gene-bod1l` - each declined to record a cell state, in writing, because no replication-stress or interstrand-crosslink-repair term existed in the 51-term cell vocabulary. It is written as a state **this** disease occupies rather than as a definition. The strongest functional evidence is pharmacological and murine: in mouse AML driven by MLL-ENL with an activated N-RAS, ATR inhibition induced chromosomal breakage and death independently of p53, and single-agent ATR or ATM inhibition had antitumoral activity in an immunocompetent allograft and in a human AML-MLL xenograft [E1]. The same state appears on the other side of the disease's life history: in human cells, replication stress activates Endonuclease G, which cleaves the MLL breakpoint cluster region and drives recombination there, with an ATM-to-RNF20 chromatin-relaxation requirement [E2]. Two further strands make the state legible in this graph's own vocabulary - the arm of the state that fails in Fanconi anaemia produces a p53/p21-dependent arrest that eliminates human HSPCs [E3], and fork protection is itself a **chromatin** function, since H3K4 methylation by SETD1A (a COMPASS-family methyltransferase, KMT2A's own family) is what licenses FANCD2's histone-chaperone activity at stalled forks [E4]. What is NOT established is that the index fusion, KMT2A::AFF1 in human infant B-lineage disease, occupies this state: nobody has measured fork speed or origin firing in a KMT2A-rearranged cell, which is the point of `exp-replication-stress-readout-kmt2a-r`.

Go deeper
Why this is a state this disease occupies, and not a textbook definition

A cell-state node in this graph has to earn its place by being a state these genes and this disease actually occupy. Three independent footholds, in decreasing order of directness:

  1. A drug-response phenotype. In mouse AML cells carrying an MLL-ENL fusion together with a constitutively active N-RAS, inhibiting ATR - the kinase that senses replication stress specifically - caused chromosomal breakage and death, and did so independently of p53, which matters because these tumours are chemoresistant through an attenuated p53 response. ATR inhibition alone reduced established tumour burden and prevented tumour growth in an immunocompetent allograft, and had activity in a xenograft of a human AML-MLL line. ATM inhibition also prolonged survival 1. A cell that dies when you block its replication-stress sensor is, operationally, a cell that was using it.
  2. A genesis phenotype. Replication stress is one of the published routes by which the MLL breakpoint cluster region breaks at all. In human cells carrying the MLLbcr hotspot in a recombination reporter, replication stress drove nuclear accumulation of Endonuclease G, EndoG-dependent cleavage of the chromosomal MLLbcr, and MLLbcr-specific recombination; breakage additionally required ATM signalling to the H2B monoubiquitin ligase RNF20 2.
  3. A dependency pattern in this project's own screen, which is NOT evidence and is recorded as such. The Fanconi-pathway ranks that prompted this beat (FANCF 24, FANCI 52, BRIP1 63, FANCD2 465 of 17,728 in the internal KMT2A-selective scan) are an internal analysis and are deliberately not cited as evidence anywhere in this entry. A dependency ranking is a hypothesis about a state; it is not a measurement of one.
The repair arm, and what its failure costs the blood compartment

The interstrand-crosslink half of this state is the half with human genetics behind it. In Fanconi anaemia patients, a profound HSPC defect is present before clinical marrow failure: in response to replicative stress and unresolved DNA damage p53 is hyperactivated and triggers a late p21(Cdkn1a)-dependent G0/G1 arrest, and p53 knockdown rescues the HSPC defect in several in vitro and in vivo models 3.

The direction of that result is the part that keeps being misread, and the five entries that wanted this node all state it correctly: FA-pathway loss imposes an arrested state, it does not maintain one. So the honest relation between this node and cell-state-quiescence is that the two are adjacent and opposite in causality - losing the ICL-repair arm pushes a cell into arrest. An edge from a Fanconi gene to quiescence would assert the wrong direction, which is why those entries declined it.

This PMID is already cited in nine entries in this graph (gene-atm, gene-bod1l, gene-brip1, gene-fancd2, gene-fancf, gene-fanci, gene-mat2a, gene-runx1, gene-usp7). It is cited here only to define what the state's failure mode looks like in a haematopoietic cell; the per-gene reasoning belongs in those entries and is not restated.

Fork protection is a chromatin function, which is why this belongs in a KMT2A graph

The reason this state is not merely borrowed DNA-repair biology is that the machinery that protects a stalled fork runs through histone methylation. Depletion of SETD1A sensitises human cells to replication stress and causes uncontrolled DNA2-dependent resection of damaged forks; the protection is mediated by SETD1A's catalysis of H3K4 methylation at replication forks, which enhances FANCD2-dependent histone-chaperone activity, and either suppressing H3K4 methylation or expressing a chaperone-defective FANCD2 mutant destabilises RAD51 nucleofilaments and causes severe fork degradation 4.

SETD1A is a COMPASS-family H3K4 methyltransferase - the same family as KMT2A. That does not mean KMT2A does this job; nobody has shown it does, and SETD1A and KMT2A are not interchangeable. What it means is that the state has a chromatin input of exactly the kind this graph is built around, which makes 'is the KMT2A fusion's chromatin programme feeding this state?' a question with a mechanism attached rather than a slogan. mech-transcription-replication-conflict-kmt2a-r holds the chain and its unverified links; exp-replication-stress-readout-kmt2a-r holds the experiment that would settle it.

How to score the state, and two ways to be fooled

Readouts that measure the state: DNA fibre or molecular combing (fork speed, inter-origin distance, restart efficiency after hydroxyurea or aphidicolin); nascent-strand degradation assays; RPA and RAD51 focus formation; pCHK1 S345 and pRPA32 S33 as ATR-pathway activation markers; 53BP1 nuclear bodies and micronuclei in the following G1; MMC/cisplatin sensitivity and chromosome-breakage analysis for the ICL arm specifically.

Fooling mode 1 - the dependency rank. A gene can rank as selectively required without the cell occupying the state, and can be pan-essential (so uninterpretable) while the state is real. Every FA gene in this graph carries its own pan-essentiality figure for that reason.

Fooling mode 2 - a deleted gene scores zero. A homozygously deleted gene has nothing for CRISPR to cut and scores near zero regardless of biology, so a screen cannot detect a loss-of-gene state from the gene's own score. That is recorded for the MTAP case in ana-the-kmt2a-arm-is-not-mtap-deleted-and-chronos-cannot-see-a-deletion and applies identically to any FA gene lost by deletion rather than mutation.

Facts and figures
ontology noteCell Ontology has no term for this state; it is a physiological condition rather than a cell identity. GO:0000724 (double-strand break repair via homologous recombination), GO:0036297 (interstrand cross-link repair) and GO:0031297 (replication fork processing) are the nearest structured vocabulary, and are process terms, not cell-state terms.
state definitionA condition in which DNA replication forks stall, slow or are actively degraded often enough that the cell's survival depends on the fork-protection and interstrand-crosslink-repair machinery (ATR/CHK1 signalling, the Fanconi anaemia core complex and ID2 module, RAD51 nucleofilament stabilisation, and fork-associated chromatin writers). It is a condition a cell is in, not a lineage it belongs to, and it is separable from quiescence: unresolved stress DRIVES arrest rather than being maintained by it.
why this node existsFive entries declined in writing to record a cell state for lack of this vocabulary term: gene-fancd2, gene-fanci, gene-fancf and gene-brip1 each carry a `declined_vocabulary` field saying there is no replication-stress or interstrand-crosslink-repair state in the graph's 51-term cell vocabulary, and gene-bod1l declined to map 'fork protection' onto cell-state-quiescence or cell-state-stem-like-self-renewal. Those refusals were correct and this node is the answer to them.
relation to cell state quiescenceADJACENT AND CAUSALLY OPPOSITE. FA-pathway loss imposes a p53/p21-dependent G0/G1 arrest (PMID 22683204); the gene does not maintain quiescence. An edge from an FA gene to cell-state-quiescence would assert the wrong direction of causation, which is why the five entries refused it.
shown for fusionPARTIAL - shown pharmacologically in MOUSE MLL-ENL AML with activated N-RAS (ATR/ATM inhibition, PMID 27625305) and shown for the MLL breakpoint cluster region in human reporter cells under replication stress (PMID 25132265). NOT shown in human KMT2A::AFF1 B-lineage disease: no fork-dynamics measurement exists in any KMT2A-rearranged cell, which is the whole point of exp-replication-stress-readout-kmt2a-r.
internal scan not cited as evidenceThe Fanconi-pathway ranks in this project's KMT2A-selective DepMap scan (FANCF 24, FANCI 52, BRIP1 63, FANCD2 465 of 17,728) are an internal analysis and are not used as a citation anywhere in this node. A dependency rank is a hypothesis about a state, not a measurement of one.
detection trapA homozygously deleted gene has nothing for CRISPR to cut and scores near zero on Chronos regardless of biology, so a dependency screen cannot detect a deletion-defined state from the deleted gene's own score. Recorded for the MTAP case in ana-the-kmt2a-arm-is-not-mtap-deleted-and-chronos-cannot-see-a-deletion; it applies identically to any FA gene lost by deletion.
operational readouts
[
  "DNA fibre / molecular combing: fork speed, inter-origin distance, restart efficiency after hydroxyurea or aphidicolin",
  "nascent-strand degradation (IdU/CldU) as the specific fork-protection readout",
  "RPA and RAD51 focus formation; pCHK1 S345, pRPA32 S33",
  "53BP1 nuclear bodies and micronuclei in the subsequent G1",
  "mitomycin C or cisplatin sensitivity plus chromosome-breakage analysis, for the interstrand-crosslink arm specifically"
]
genes in this graph that serve this state
[
  {
    "gene": "gene-fancd2",
    "role": "ID2 module effector; also the histone chaperone at protected forks (PMID 29937342)"
  },
  {
    "gene": "gene-fanci",
    "role": "obligate ID2 heterodimer partner of FANCD2"
  },
  {
    "gene": "gene-fancf",
    "role": "FA core complex adaptor"
  },
  {
    "gene": "gene-brip1",
    "role": "FANCJ helicase, downstream effector; note the BACH1 alias trap logged in data/symbol-collisions.json"
  },
  {
    "gene": "gene-bod1l",
    "role": "fork protection; suppresses DNA2-dependent resection by stabilising RAD51 (PMID 26166705)"
  },
  {
    "gene": "gene-atr",
    "role": "the sensor kinase whose inhibition kills mouse MLL-ENL AML (PMID 27625305)"
  },
  {
    "gene": "gene-atm",
    "role": "double-strand-break sensor; also required for MLLbcr breakage via RNF20 (PMID 25132265)"
  },
  {
    "gene": "gene-setd1a",
    "role": "COMPASS-family H3K4 methyltransferase that licenses FANCD2 chaperone activity at forks (PMID 29937342)"
  }
]
How this could bite us back
  • ATR and ATM inhibitors are not selective for leukaemia: every proliferating tissue in an infant - gut crypt, marrow, thymus, growing bone - is under replication stress, and the therapeutic window in the cited work is measured in mice with intact marrow, not in a child after induction chemotherapy.
  • Haematopoietic stem cells are the cell type in which failure of this state's repair arm is most catastrophic: FA patients lose HSPCs through exactly this pathway (PMID 22683204). An agent that pushes leukaemic cells over a replication-stress threshold has no obvious reason to spare the normal HSPC pool, and the consequence would be marrow failure that is indistinguishable from treatment-refractory disease at the bedside.
  • The same state is implicated in CREATING MLL rearrangements (PMID 25132265). An agent that raises replication stress in a child's haematopoietic compartment is, by the mechanism this entry records, an agent that could generate the class of lesion being treated - which is the therapy-related AML problem this graph already tracks for topoisomerase-II poisons, arriving by a second route.
  • Because p53 is attenuated in MLL-rearranged AML, the usual death route for replication catastrophe is impaired; the cited ATR result works BECAUSE it is p53-independent. That also means the standard pharmacodynamic markers of a successful DNA-damage response (p53 stabilisation, p21 induction) will not report target engagement in this disease.
  • Fork protection is supplied redundantly - SETD1A/FANCD2, BOD1L, RAD51 stabilisation, BLM and FBH1 restraint are partly interchangeable at the same endpoint - so single-gene inhibition of this state is a strong candidate for compensation, and a combination is the likely requirement.
Still open
  1. Do KMT2A-rearranged cells actually replicate their DNA differently? No fork speed, inter-origin distance or restart measurement exists in any KMT2A-rearranged cell, human or mouse. The entire Fanconi dependency argument in this graph is inferred from a screen ranking; exp-replication-stress-readout-kmt2a-r is the DNA-fibre experiment that would confirm or kill it, and it has not been done.
  2. Is the ATR sensitivity in the MLL-ENL model a property of the fusion or of the co-expressed activated N-RAS? Activated RAS alone causes hyper-replication and fork-progression defects (PMID 17136094), so the model confounds the two. The discriminating experiment is ATR inhibition in a RAS-wild-type KMT2A-rearranged line.
  3. Does the index fusion's own chromatin programme feed this state? SETD1A-deposited H3K4 methylation protects forks through FANCD2 (PMID 29937342) and KMT2A is in the same methyltransferase family, but no experiment has asked whether KMT2A or a KMT2A fusion methylates at forks. A KMT2A ChIP over nascent DNA (iPOND-ChIP) in a t(4;11) line is the direct test.
  4. Is the replication-stress arm and the interstrand-crosslink arm one state or two in this disease? The FA genes in this graph rank high as a group, but ICL repair and fork protection are separable functions with separable readouts (MMC breakage versus nascent-strand degradation), and nothing here has separated them.
Every source behind this entry (4)
  1. E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    In mouse AML driven by an MLL-ENL fusion with a constitutively active N-RAS, inhibition of the replication-stress sensor kinase ATR induced chromosomal breakage and cell death independently of p53, and ATR inhibition as a single agent reduced tumour burden after establishment and prevented tumour growth in an immunocompetent allograft model and in a xenograft of a human AML-MLL cell line; ATM inhibition also promoted survival of the AML(MLL) mice.

    Targeting the kinase activities of ATR and ATM exhibits antitumoral activity in mouse models of MLL-rearranged AML. · Morgado-Palacin I, Day A, Murga M, Lafarga V, Anton ME, Tubbs A, et al., Fernandez-Capetillo O · Sci Signal 2016
    PMID 27625305DOI
    What this evidence cannot show

    Murine MLL-ENL myeloid disease, not human infant B-lineage KMT2A::AFF1, and the model carries a constitutively active N-RAS as a second lesion. The experiment measures sensitivity to blocking a stress-response kinase; it does not measure fork dynamics, so it establishes that the pathway is load-bearing rather than that the cells were quantifiably under more replication stress than a comparator. Read: abstract only. Already cited in this graph in gene-atr, mech-transcription-replication-conflict-kmt2a-r and exp-replication-stress-readout-kmt2a-r; cited here for the state, not for the per-gene conclusion.

    Evidence that argues against this (1)
    • The second lesion may be doing the work. Activated RAS is itself a documented driver of replication stress: in normal human cells, expression of activated H-RasV12 triggers a hyper-replicative phase with augmented numbers of active replicons and altered fork progression, and the resulting DNA-damage response is what enforces oncogene-induced senescence. An MLL-ENL model that also carries constitutively active N-RAS therefore cannot separate a fusion-driven replication-stress state from a RAS-driven one, and the ATR sensitivity may be a property of the RAS allele. Read: abstract only. Di Micco R et al. 2006
  2. E2moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

    In human cells carrying the MLL breakpoint cluster hotspot in an EGFP-based recombination reporter, replication stress drove nuclear accumulation of Endonuclease G and EndoG-dependent cleavage of the chromosomal MLLbcr locus with MLLbcr-specific recombination, and MLLbcr breakage additionally required ATM signalling to the H2B monoubiquitin ligase RNF20.

    Endonuclease G initiates DNA rearrangements at the MLL breakpoint cluster upon replication stress. · Gole B, Baumann C, Mian E, Ireno CI, Wiesmueller L · Oncogene 2015
    PMID 25132265DOI
    What this evidence cannot show

    A reporter-substrate system in cell lines, not primary fetal haematopoietic cells and not a patient sample; the readout is recombination of an introduced substrate plus cleavage of the endogenous locus, not the generation of an actual KMT2A fusion. It establishes that replication stress can break this locus, not that it is how any given patient's fusion arose - the competing route in this graph is topoisomerase-II-mediated cleavage. Read: abstract only.

    Evidence that argues against this (1)
    • Damage to the haematopoietic stem-cell genome in vivo has been characterised as aldehyde-driven double-strand breakage repaired by microhomology-mediated end joining, with deletions and rearrangements read out by single-HSC transplantation plus whole-genome sequencing, and with p53 deletion fully rescuing survival of aldehyde-stressed stem cells without changing the pattern or intensity of instability. That is a distinct initiating chemistry from stalled-fork replication stress, and it means an MLL-locus rearrangement in a real stem cell need not have a replication-stress origin. Read: abstract only. Garaycoechea JI et al. 2018
  3. E3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    In Fanconi anaemia patients, haematopoietic stem and progenitor cells carry a profound defect that is present before the onset of clinical bone marrow failure: in response to replicative stress and unresolved DNA damage p53 is hyperactivated and triggers a late p21(Cdkn1a)-dependent G0/G1 cell-cycle arrest, and p53 knockdown rescues the HSPC defects in several in vitro and in vivo models including human FA or FA-like cells.

    Bone marrow failure in Fanconi anemia is triggered by an exacerbated p53/p21 DNA damage response that impairs hematopoietic stem and progenitor cells. · Ceccaldi R, Parmar K, Mouly E, Delord M, Kim JM, Regairaz M, et al., Soulier J · Cell Stem Cell 2012
    PMID 22683204DOI
    What this evidence cannot show

    Fanconi anaemia patients and FA-like models, which are fusion-free: nothing here is KMT2A-rearranged, leukaemic or B-lineage. It describes what happens when the ICL-repair arm of this state FAILS constitutionally over years, which is a different question from whether a leukaemic cell transiently occupies the state. Read: abstract only. This PMID is already cited in nine entries here, so its per-gene conclusions are written down elsewhere and are not repeated in this node.

    Evidence that argues against this (1)
    • The same haematopoietic phenotype has a rival aetiology that does not require a replication-stress reading. Endogenous and alcohol-derived acetaldehyde damages chromosomes and mutates haematopoietic stem cells directly, and p53 deletion fully rescues the survival of aldehyde-stressed stem cells while leaving the genome instability itself unchanged - so 'p53-dependent HSPC attrition' is a shared downstream readout of at least two different upstream chemistries, and observing it does not identify which one a cell is under. Read: abstract only. Garaycoechea JI et al. 2018
  4. E4strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    In human cells, depletion of the COMPASS-family H3K4 methyltransferase SETD1A sensitises cells to replication stress and causes uncontrolled DNA2-dependent resection of damaged replication forks; the protection is mediated by SETD1A-catalysed H3K4 methylation at replication forks, which enhances FANCD2-dependent histone-chaperone activity, and suppressing H3K4 methylation or expressing a chaperone-defective FANCD2 mutant destabilises RAD51 nucleofilaments and causes severe nucleolytic fork degradation.

    Histone Methylation by SETD1A Protects Nascent DNA through the Nucleosome Chaperone Activity of FANCD2. · Higgs MR, Sato K, Reynolds JJ, Begum S, Bayley R, Goula A, et al., Stewart GS · Mol Cell 2018
    PMID 29937342DOI
    What this evidence cannot show

    SETD1A, not KMT2A, and human non-haematopoietic cell lines under induced replication stress. The COMPASS-family link is an argument for why this state is mechanistically adjacent to this graph's subject matter; it is not evidence that KMT2A or a KMT2A fusion performs the same function, and no such experiment exists. Read: abstract only.

    Evidence that argues against this (1)
    • The fork-protection function this state is invoked for can be supplied by a factor with no chromatin role at all, which weakens the inference that H3K4 methylation is the controlling input. BOD1L is required to suppress deleterious resection of stressed forks by stabilising RAD51 at them, and blocking DNA2-dependent resection or depleting BLM and FBH1 suppresses both the catastrophic fork processing and the chromosomal damage in BOD1L-deficient cells - the same endpoint, the same nuclease, no methyltransferase. Read: abstract only. Higgs MR et al. 2015
Connected to