Oncofusion

start / graph / exp-c-circle-and-telomere-fish-across-the-thirteen-b-all-lines

Experimentspeculative confidence 1 of 5: Our own inference or analogy. Not shown to be true.Q172-daxx-atrx-h33-axis

Assay telomere maintenance across the thirteen B-lineage ALL DepMap lines, and see whether it predicts the DAXX dependency split

also called the C-circle experiment · thirteen lines, one assay

Grow thirteen leukaemia cell lines that are already sitting in freezers worldwide, and run one cheap test on each that says how they keep their chromosome ends intact. If the six cells that need DAXX are also the six with the unusual telomeres, we have an explanation; if not, we have eliminated the only one we had.

The same thing, for a scientist

The named lines are already the exact panel DepMap screened, so the dependency readout exists and does not need repeating. The experiment adds one axis - telomere maintenance state - and asks whether it partitions the panel the same way DAXX Chronos does. **Primary readout: the C-circle assay**, which is the standard quantitative ALT marker; **secondary: telomere FISH** for the heterogeneous ultrabright-foci appearance, and telomere restriction fragment Southern for length distribution. Controls: U2OS (ALT-positive) and HeLa (telomerase-positive, ALT-negative), run in the same batch. **Design point that matters.** The hypothesis being tested is not "these lines are ALT" - acute leukaemia is overwhelmingly telomerase-positive and that would be a surprising result. It is the weaker and more plausible "ATRX/DAXX-dependent H3.3 deposition is load-bearing for telomere integrity in these six lines", which predicts a **graded** C-circle signal and heterogeneous telomere lengths rather than frank ALT. So the analysis must be pre-specified as a **rank correlation between C-circle signal and DAXX Chronos across all thirteen lines**, not as a binary ALT call, or the experiment can only return a negative. **Cost tier: low.** Thirteen commercially available lines, one assay, no animals, no primary material, no sequencing. Weeks, not months.

Go deeper
The panel, with the number the experiment has to explain
lineCVCLfusionDAXX Chronospredicted group
HB11;19CVCL_8227KMT2A::MLLT1-0.979deep
SEMCVCL_0095KMT2A::AFF1-0.948deep
RS4;11CVCL_0093KMT2A::AFF1-0.821deep
RCH-ACVCVCL_1851TCF3::PBX1-0.802deep
SEM-K2CVCL_S906KMT2A::AFF1 (SEM derivative)-0.798deep
YCUB-2CVCL_A082TCF3::HLF-0.792deep
NALM-6CVCL_0092IGH::DUX4-0.319shallow
RehCVCL_1650ETV6::RUNX1-0.309shallow
JM1CVCL_F242--0.245shallow
697CVCL_0079TCF3::PBX1-0.232shallow
TANOUECVCL_1852--0.203shallow
P30/OHKCVCL_1631MEF2D::HNRNPUL1-0.138shallow
NALM-16CVCL_1834--0.097shallow

The single most informative pair is RCH-ACV against 697. Both carry TCF3::PBX1, both are paediatric, and they are 0.57 apart in DAXX dependency. If telomere state separates those two, the hypothesis is in good shape on one comparison. If it does not, nothing else in the panel will rescue it.

Include SEM and SEM-K2 both, and treat their agreement as a technical replicate rather than as two data points - they are one donor, and a discordant result between a line and its own derivative would mean the assay, not the biology, is doing the work.

Pre-specified analysis, written before the data exist
  • Primary: Spearman correlation between C-circle signal (normalised to U2OS) and DAXX Chronos across all thirteen lines. Pre-specified threshold for support: rho more negative than -0.6 with a permutation p below 0.05 over 20,000 label shuffles. n = 13, so this is not well powered and a null result is not strong evidence of absence; that is stated now rather than discovered later.
  • Secondary: the RCH-ACV versus 697 contrast, reported on its own whatever the primary shows.
  • Pre-specified negative: if no line exceeds 10% of the U2OS C-circle signal, report the hypothesis REFUTED and say so in hyp-daxx-dependence-in-kmt2a-r-b-all-is-a-telomere-phenotype's falsifier field, in place, per this project's rule 96.
  • Do not dichotomise into ALT-positive and ALT-negative and run a Fisher test. The hypothesis predicts a gradient; a binary call throws away the only variation the experiment generates and would make a supporting result look like a null.
  • Blind the assay to the DAXX values. The person running the C-circles should not know which lines are in the deep group, because both readouts have a subjective component.
The free version, which should be done first

These thirteen are among the most-used B-ALL lines in existence and telomere measurements on some of them may already be published or deposited. Before any bench work:

  1. Search the literature for telomere length, telomerase activity or C-circle data on each line by name and by CVCL accession, remembering that RS4;11 is also written RS4-11, RS4:11 and RS411 and that SEM-K2 is also SEMK2.
  2. Search GEO and ArrayExpress for the lines by name; telomere content is recoverable from WGS read pileups and several of these lines are whole-genome sequenced in CCLE.
  3. Check whether CCLE's own data include a telomere-length estimate - it is derivable from the existing WGS and would answer the question for free.

An hour of searching could support, eliminate or de-prioritise the experiment before anyone thaws a vial.

Facts and figures
designThirteen B-lineage ALL cell lines (the complete DepMap B-ALL panel: SEM, SEM-K2, RS4;11, HB11;19, RCH-ACV, YCUB-2, NALM-6, Reh, JM1, 697, TANOUE, P30/OHK, NALM-16) plus U2OS as ALT-positive control and HeLa as telomerase-positive ALT-negative control, all cultured and harvested in one batch, STR-confirmed on receipt, assayed blind to the DAXX dependency values.
readoutPrimary: C-circle assay, phi29-dependent rolling-circle amplification of C-rich telomeric circles, signal normalised to U2OS. Secondary: telomere FISH scored for ultrabright heterogeneous foci per nucleus, and telomere restriction fragment Southern for the length distribution. Tertiary, if the primary supports the hypothesis: DAXX shRNA or degron depletion in one deep and one shallow line with a time course of telomere dysfunction-induced foci (53BP1 at telomeres) against HOXA9/MEIS1 qPCR, to order the two consequences.
controlsU2OS (ALT-positive) and HeLa (ALT-negative telomerase-positive) in the same batch; phi29-minus reaction for every sample; SEM and SEM-K2 as a same-donor pair whose agreement reports assay reproducibility rather than biology; blinding of the operator to the Chronos values.
cost tierlow
timeline6-10 weeks including line acquisition and STR confirmation; the literature and GEO sweep that should precede it is one day.
n lines13
prespecified primary analysisSpearman rank correlation between normalised C-circle signal and DAXX Chronos across all 13 lines; support threshold rho < -0.6 with permutation p < 0.05 over 20,000 label shuffles. Underpowered at n=13 by design and stated as such in advance.
prespecified negativeNo line exceeding 10% of the U2OS C-circle signal is reported as REFUTED, written into the falsifier field of hyp-daxx-dependence-in-kmt2a-r-b-all-is-a-telomere-phenotype rather than into a new entry.
statusplanned
do first at no cost
[
  "Literature search for telomere length, telomerase activity or C-circle data on each of the thirteen lines by name and CVCL accession, including the RS4;11 / RS4-11 / RS4:11 / RS411 and SEMK2 / SEM-K2 spellings",
  "GEO and ArrayExpress search for the lines by name",
  "Check whether CCLE whole-genome sequencing supports a telomere-length estimate for these lines, which would answer the primary question with no wet work"
]
How this could bite us back
  • A positive result makes DAXX a WORSE target, not a better one: it would say that removing DAXX in these cells attacks telomere maintenance, which is precisely the selective pressure that produces ALT-positive escape in other tumours.
  • The panel contains no infant-derived line and no primary patient material, so a clean positive result still would not be about the index disease. Any follow-on must include primary infant KMT2A::AFF1 blasts, where the C-circle assay is harder and the cell numbers are limiting.
Still open
  1. Does CCLE whole-genome sequencing already support a telomere-length estimate for these thirteen lines? If it does, the primary question is answerable today with no bench work and this entry should be replaced by an analysis.
  2. Is there a published telomerase activity (TRAP) measurement for RS4;11, SEM or Reh? These lines are old and heavily characterised, and a TRAP result would not decide the hypothesis but would set the prior.
  3. If the correlation holds, is it causal? A telomere state that correlates with DAXX dependence across thirteen lines still needs the depletion time-course to show that telomere damage precedes the growth defect, and that experiment is not designed here.
Every source behind this entry (3)
  1. E1established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper

    The C-circle assay is the established quantitative marker of alternative lengthening of telomeres and ATRX or DAXX loss is the recurrent genetic lesion associated with it in human tumours, with 61% of pancreatic neuroendocrine tumours ALT-positive and all of those carrying an ATRX or DAXX mutation or nuclear protein loss.

    Altered telomeres in tumors with ATRX and DAXX mutations. · Heaphy CM, de Wilde RF, Jiao Y, et al. · Science 2011
    PMID 21719641DOI
    What this evidence cannot show

    Abstract read from efetch 2026-09-24. This paper uses telomere FISH morphology rather than the C-circle assay; the C-circle assay is named here as the modern quantitative standard and that choice is this entry's, not the paper's. The association established is in solid tumours, and no leukaemia is examined.

    Evidence that argues against this (1)
    • Choosing the C-circle assay as primary embeds an assumption: that whatever makes these lines DAXX-dependent produces extrachromosomal C-rich circles. If the state is a partial dependence on ATRX/DAXX-mediated H3.3 deposition at telomeres WITHOUT recombination-based lengthening, C-circles will be flat and telomere FISH heterogeneity may be the only readout that moves. Running FISH and TRF alongside rather than as afterthoughts is not redundancy; it is the difference between a discriminating experiment and one that can only fail. Goldberg AD et al. 2010
  2. E2established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper

    Atrx is required for Hira-independent localisation of histone H3.3 at telomeres and for repression of telomeric RNA, and Atrx and Daxx associate with H3.3 independently of Hira, establishing that the telomeric arm of H3.3 deposition is biochemically separable from the gene-body arm.

    Distinct factors control histone variant H3.3 localization at specific genomic regions. · Goldberg AD, Banaszynski LA, Noh KM, et al. · Cell 2010
    PMID 20211137DOI
    What this evidence cannot show

    Abstract read from efetch 2026-09-24. Mouse embryonic stem cells, not leukaemia and not human. This is the founding separability result and is cited here for the premise that a telomeric readout is the right place to look for an ATRX/DAXX phenotype, not for anything about leukaemia. Already cited in `gene-atrx` as E1; re-fetched here rather than copied.

    Evidence that argues against this (1)
    • If the telomeric and gene-body arms are separable, a DAXX dependency in a leukaemic blast could equally well be the gene-body arm - and this graph's own reasoning says HIRA, not DAXX, is the gene-body chaperone. But HIRA's lymphoid difference is +0.032, i.e. slightly LESS required in the index arm, so the gene-body reading has no support in the dependency data. That asymmetry is why this experiment targets telomeres, and it is an argument from elimination rather than from evidence. Oncofusion knowledge graph (internal analysis) 2026
  3. E3moderate confidence 3 of 5: One strong study, or several smaller consistent ones.dataset

    The thirteen B-lineage ALL lines that constitute the panel, their Cellosaurus accessions, their annotated fusions and their DAXX Chronos values are all recorded, and the six-versus-seven split the experiment must explain is separated by a 0.473 gap containing no line.

    What this evidence cannot show

    Chronos values are from Open Targets dataVersion 26/09 fetched 2026-09-24 and will shift with each quarterly release, so the panel should be re-pulled at the time the experiment is run rather than taken from this entry. Line identity should be confirmed by STR profiling on receipt; several of these lines have documented misidentification histories in the wider literature and Cellosaurus is the authority on which.

    Evidence that argues against this (1)
    • Thirteen lines is the whole population available, not a sample, which is a strength for describing DepMap and a weakness for inference: there is no held-out set, so a telomere correlation found here cannot be validated except by adding lines outside DepMap, where the DAXX dependency would itself have to be re-measured. Any positive result from this experiment is a hypothesis-generating correlation over thirteen points and should be written as one. Oncofusion knowledge graph (internal analysis) 2026
Connected to