LIN28B has THREE published signs in myeloid leukaemia - oncogene, tumour suppressor, and differentiation agent - and the design quality runs opposite to the direction the index-fusion story needs
also called the LIN28B sign problem · LIN28B counter-evidence dossier · is LIN28B an oncogene or a tumour suppressor
Four careful papers disagree about whether LIN28B helps leukaemia or fights it. One says blocking LIN28B kills leukaemia cells; two say adding LIN28B makes leukaemia grow up and stop; one says leukaemia cannot even start without it. This entry lays out which of them is best designed, because the answer is not the one that supports the story this project is most excited about.
The same thing, for a scientist
This graph's LIN28B entry declared one counter-evidence paper **from its title alone**. Reading the literature properly turns up **four studies with three distinct signs**, and the strongest designs point away from LIN28B being pro-leukaemic. **(A) Oncogene.** Knocking down LIN28B in AML cell lines causes cell-cycle arrest and loss of proliferation and colony formation via let-7a de-repression, with IGF2BP1 downstream; LIN28B overexpression increases tumorigenicity in vivo [E1]. **(B) Tumour suppressor at initiation.** In inducible MLL-ENL mice, LIN28B *suppressed* AML initiation rather than accelerating it, and initiation peaked neonatally rather than fetally [E2]; mechanistically, LIN28B binds MYBBP1A mRNA to raise the MYB co-repressor and compromise LIC activity, with MYBBP1A overexpression phenocopying and MYBBP1A perturbation abolishing the effect [E3]. **(C) Differentiation agent in established disease.** In fully transformed MLL::AF9 AML, enforced LIN28B extended survival with **80% of secondary recipients alive beyond four months**, induced partial monocytic differentiation by day 6, and severely depleted colony-forming potential in five independent AMLs **without increasing apoptosis** [E4]. **(D) Required for initiation, in the index fusion.** Human fetal liver KMT2A::AFF1 [E5]. **And the same study that gives (C) reports the necessity arm as NEGATIVE**: conditional *Lin28b* deletion in fetal progenitors did not accelerate MLL::ENL AML - LIN28B is not the effector of fetal protection either [E4].
The four studies graded by design, not by conclusion
| design | perturbation | system | LIN28B is... | design strength | |
|---|---|---|---|---|---|
| A Zhou 2017 1 | knockdown/overexpression in cell lines + xenograft | both | unselected AML lines, no fusion named | oncogene | weakest. Cell lines, no fusion, and an RBP modulated by transfection in cells that may not express it physiologically |
| B Okeyo-Owuor 2019 2 | inducible transgenic, four developmental windows | gain | mouse MLL-ENL | tumour suppressor at initiation | strong design; transgene and mouse |
| B' Eldeeb 2023 3 | dual transgenic + mechanism, sufficiency AND necessity on MYBBP1A | gain | mouse MLL-ENL | tumour suppressor, mechanism traced | strongest mechanism; transgene dose is supraphysiological |
| C Li 2024 4 | Tet-ON/Tet-OFF, secondary transplants, 5 independent AMLs | gain, and fetal loss | mouse MLL::AF9 and MLL::ENL | differentiation agent; and NOT required for fetal protection | strongest genetics; mouse, transgene, supraphysiological by the authors' own statement |
| D Ling 2026 5 | endogenous CRISPR translocation, LIN28B-null | loss | human fetal liver CD34+, KMT2A::AFF1 | required for initiation | most relevant system by far; read here from the ABSTRACT ONLY, paywalled, controls unread |
The uncomfortable summary. The study whose system is most relevant to this project (D) is the one this graph can read least of. The studies whose genetics are cleanest (B', C) are in the wrong species, the wrong lineage and the wrong partner, and they point the other way. A reader who weighted only by design would conclude LIN28B is a myeloid tumour suppressor; a reader who weighted only by relevance would conclude it is required for infant B-ALL initiation. Both readings are defensible and this entry refuses to pick.
The patient-expression datum that does real work
Li et al. report, from re-analysis of external data, that most human paediatric AML including MLL-rearranged AML do not express LIN28B, and express several let-7 family miRNAs at high levels 4.
This matters more than its one sentence suggests, because it does two things at once:
- It undercuts study A. If patient MLLr AML does not express LIN28B, a LIN28B dependency in unselected AML lines is a property of the lines, not of the disease. That is the single best argument for setting the oncogene reading aside.
- It supports the lineage split. Infant ALL expresses LIN28B (Ling et al.); paediatric AML largely does not. The expression pattern itself, in human patients, follows lineage rather than age.
What it does NOT do is rescue the index-fusion story, because it says nothing about whether LIN28B is required where it is expressed. It removes a contradiction rather than adding support.
What this project got wrong, named
gene-lin28b (Q132, cycle 106) cites PMID 38321106 as counter-evidence and states in its own caveats: 'TITLE ONLY. This record was found by a PubMed title search and its abstract was not retrieved.' That was honest labelling, and it was still the wrong depth for the entry's most important counter-evidence.
Reading the full text (PMC10912017 is open access) changed three things:
- The title says 'dispensable for fetal leukemia suppression', which reads as a null. The paper's positive result is much stronger than its title - 80% survival beyond four months on enforced LIN28B, in five independent AMLs.
- It revealed the MYBBP1A mechanism and the Eldeeb paper behind it, neither of which was anywhere in this corpus.
- It revealed the human patient expression datum above, which is the most decision-relevant sentence in the whole dossier.
The generalisable point: a title-only counter-evidence citation is a placeholder, and an open-access full text is fifteen seconds away. This entry exists partly as the corrected version of that field.
| negative result that travels least | Conditional Lin28b deletion in fetal progenitors did NOT accelerate MLL::ENL AML initiation. LIN28B is therefore not the effector of fetal protection even in the model that established fetal protection - the gain-of-function arm is strong and the loss-of-function arm is empty. |
|---|---|
| human patient expression | Most paediatric AML including MLLr AML do NOT express LIN28B and express let-7 family miRNAs highly (Li 2024, re-analysis). Infant ALL DOES express LIN28B (Ling 2026). Expression follows lineage, in human patients. |
| what this entry corrects | gene-lin28b cited PMID 38321106 from its TITLE ONLY, by its own admission. The full text is open access (PMC10912017), and reading it surfaced the MYBBP1A mechanism, the Eldeeb paper, the 80%-survival result, and the human patient expression datum - none of which were in this corpus. |
| shown for fusion | KMT2A::MLLT1 and KMT2A::MLLT3 (mouse, tumour-suppressive / differentiating); KMT2A::AFF1 (human, required for initiation); no fusion named for the oncogene reading. |
signs reported — the numbers behind the chart
[
{
"sign": "oncogene",
"study": "Zhou 2017, PMID 28693523",
"system": "unselected AML cell lines + xenograft, no fusion named",
"mechanism": "let-7a de-repression, IGF2BP1 downstream",
"design_rank": 4
},
{
"sign": "tumour suppressor at initiation",
"study": "Okeyo-Owuor 2019, PMID 31405949; Eldeeb 2023, PMID 36763502",
"system": "mouse transgenic MLL-ENL",
"mechanism": "MYBBP1A mRNA binding -> MYB repression -> LIC activity down",
"design_rank": 2
},
{
"sign": "differentiation agent in established disease",
"study": "Li 2024, PMID 38321106",
"system": "mouse MLL::AF9, Tet-ON/OFF, secondary transplants",
"mechanism": "monocytic differentiation, clonogenic collapse, no apoptosis increase",
"design_rank": 1
},
{
"sign": "required for initiation",
"study": "Ling 2026, PMID 42213620",
"system": "human fetal liver CD34+, endogenous CRISPR KMT2A::AFF1",
"mechanism": "stabilisation of B-lymphoid, epigenetic, cell-cycle and anti-apoptotic transcripts",
"design_rank": "most RELEVANT system, least readable - abstract only"
}
]read depth by study
{
"28693523": "abstract",
"31405949": "abstract",
"36763502": "abstract",
"38321106": "FULL TEXT (PMC10912017)",
"42213620": "abstract only, paywalled, not in PMC"
}- Two of the four studies show that ADDING LIN28B is anti-leukaemic in myeloid disease. If that generalises even partly, a LIN28B inhibitor is the wrong-signed drug for a substantial fraction of KMT2A-rearranged patients.
- Lineage switch is documented in KMT2A-rearranged leukaemia under lineage-directed immunotherapy. A patient whose disease switches from B to myeloid could move from the arm where LIN28B inhibition helps to the arm where it harms, mid-treatment.
- The best-read study in this dossier is the one that argues against the project's preferred direction, and the least-read is the one that supports it. That asymmetry should be treated as a bias in the graph's current position, not as a coincidence.
- If the differentiation effect is real in humans, the intervention is LIN28B RESTORATION - a gene-delivery problem in an infant, not a small-molecule problem, with an entirely different development path and risk profile.
- Does enforced LIN28B differentiate a human KMT2A::AFF1 infant ALL PDX, as it differentiates mouse MLL::AF9 AML? If it does, LIN28B is a differentiation agent in the index disease too and the initiation requirement is a separate, earlier phenomenon - which would reverse the therapeutic direction entirely, from inhibiting LIN28B to supplying it.
- Is LIN28B expressed in the AML lines Zhou et al. used, at physiological levels? If not, the oncogene reading can be set aside on design grounds alone and the dossier collapses from three signs to two.
- Does the supraphysiological transgene dose explain the myeloid tumour-suppressive effect? A single-copy endogenous-level knock-in would settle it and is the cheapest way to reconcile the dossier.
- What is the LIN28B expression level in diagnostic infant KMT2A::AFF1 blasts, in absolute terms and relative to fetal liver CD34+? Everything in this dossier is about presence or absence, and a dose-dependent sign reversal would explain all four papers at once.
- E1moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper
Targeting LIN28B in AML cells resulted in cell-cycle arrest and inhibition of proliferation and colony formation, induced by de-repression of let-7a; overexpression of LIN28B promoted cell proliferation; IGF2BP1 was confirmed as a downstream target of LIN28B via let-7 in AML; ectopic LIN28B increased tumorigenicity in vivo while silencing it slowed xenograft growth.
Inhibition of LIN28B impairs leukemia cell growth and metabolism in acute myeloid leukemia. · Zhou J, Bi C, Ching YQ, Chooi JY, Lu X, Quah JY, Toh SH, Chan ZL, Tan TZ, Chong PS, Chng WJ · J Hematol Oncol 2017What this evidence cannot show
Abstract read; full text not opened. AML CELL LINES plus a xenograft; NO FUSION IS NAMED anywhere in the abstract, so this cannot be counted as a KMT2A-rearranged result. LIN28B was modulated by transfection in lines whose physiological LIN28B status is not stated. This is the weakest design of the four and gives the sign that most flatters a pro-leukaemic reading.
Evidence that argues against this (1)
- Most human paediatric AML, including MLL-rearranged AML, do not express LIN28B and express let-7 family miRNAs highly - so a LIN28B dependency in AML lines is hard to reconcile with patient biology and most likely reflects the lines. Leukemia 2024
- E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
In inducible transgenic MLL-ENL mice, AML initiation efficiency changed with age and peaked during neonatal rather than fetal development; LIN28B, tested as a candidate accelerator, suppressed AML initiation rather than accelerating it, which the authors propose explains why MLL rearrangements arise before birth but transformation usually follows birth as LIN28B levels decline.
The efficiency of murine MLL-ENL-driven leukemia initiation changes with age and peaks during neonatal development. · Okeyo-Owuor T, Li Y, Patel RM, Yang W, Casey EB, Cluster AS, Porter SN, Bryder D, Magee JA · Blood Adv 2019What this evidence cannot show
Abstract read; full text not opened. Mouse, transgene, myeloid. Transplantation stress mitigated the developmental barrier per the abstract, meaning the age effect is not robust to handling - a limitation the authors state themselves.
Evidence that argues against this (1)
- The same abstract reports that transplantation stress mitigated the developmental barrier, so the age effect on initiation efficiency is not robust to handling. A developmental window that disappears under transplantation is a weak foundation for a claim about human disease timing. Leukemia 2024
- E3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
In a dual-transgenic mouse co-expressing MLL-ENL and LIN28B, LIN28B impedes MLL-rearranged AML progression through compromised leukaemia-initiating-cell activity and suppression of MYB signalling; LIN28B directly binds MYBBP1A mRNA, elevating this MYB co-repressor; MYBBP1A overexpression phenocopies the tumour-suppressor effect and MYBBP1A perturbation omits it.
A fetal tumor suppressor axis abrogates MLL-fusion-driven acute myeloid leukemia. · Eldeeb M, Yuan O, Guzzi N, Thi Ngoc PC, Konturek-Ciesla A, Kristiansen TA, Muthukumar S, Magee J, Bellodi C, Yuan J, Bryder D · Cell Rep 2023What this evidence cannot show
Abstract read; full text not opened. Mouse, transgene at supraphysiological LIN28B levels per the companion paper, myeloid, MLLT1 partner. The abstract itself calls the influence of prenatal signals on leukaemogenesis 'controversial', which is a useful admission from inside the field.
Evidence that argues against this (1)
- The companion study from the same collaboration found the necessity arm EMPTY - conditional Lin28b deletion in fetal progenitors did not accelerate MLL::ENL AML. A mechanism traced with sufficiency and necessity for MYBBP1A sits inside a model where LIN28B itself is not necessary for the protective phenotype, which limits how much the MYBBP1A chain can explain. Leukemia 2024
- E4strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
Conditional Lin28b deletion in fetal progenitors of Tet-OFF MLL::ENL mice did not accelerate AML initiation or shorten survival, so LIN28B is not the sole effector of fetal leukaemia suppression; enforced LIN28B in established MLL::AF9 AML significantly extended survival with 80% of secondary recipients surviving beyond 4 months and GFP+ AML cells undetectable in the marrow of most survivors, induced partial monocytic differentiation by 6 days with extensively differentiated colony morphologies, and severely depleted colony-forming potential in 5 biologically independent AMLs without increasing programmed cell death; and most human paediatric AML including MLLr AML do not express LIN28B while expressing several let-7 family miRNAs highly.
LIN28B promotes differentiation of fully transformed AML cells but is dispensable for fetal leukemia suppression. · Li Y, Mendoza-Castrejon J, Patel RM, Casey EB, Denby E, Bryder D, Magee JA · Leukemia 2024What this evidence cannot show
FULL TEXT READ (PMC10912017), which is the point of this entry - the graph previously cited this paper from its TITLE alone. It is a Letter, so compact, and many of the quantitative claims rest on supplementary figures not reproduced in the retrieved text stream. Mouse, transgenes, myeloid. The authors flag two limits themselves: LIN28B levels exceeded physiological levels, and the fetal-deletion experiment ended at ~6 months. The human AML expression claim is a re-analysis of external data, not a primary measurement.
Evidence that argues against this (1)
- None of this is in the index fusion, the index lineage, the index species or with an endogenous lesion, and the human index-fusion study reports the opposite requirement. Blood 2026
- E5strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
Human fetal liver CD34+ cells fail to transform upon induction of the KMT2A::AFF1 translocation in the absence of LIN28B, and LIN28B knockdown removes the proliferative advantage of established CRISPR-KMT2A::AFF1 leukaemias in vitro and in vivo.
The fetal specific gene LIN28B is essential for human fetal B-lymphopoiesis and initiation of KMT2A::AFF1 infant leukemia. · Ling RE, Welsh AM, Hamer L, Jackson T, Neil E, Elliott N, et al. · Blood 2026What this evidence cannot show
ABSTRACT ONLY and paywalled - Europe PMC returns isOpenAccess N, inEPMC N, inPMC N, verified live 2026-09-23. This is the most relevant study in the dossier and the one this graph can inspect least. The asymmetry of read depth across the four studies is itself a bias in how this entry weighs them, and it is stated rather than corrected.
Evidence that argues against this (1)
- Read depth is doing hidden work in this entry's grading. The study ranked most relevant is the only one read at abstract depth behind a paywall, so its design cannot actually be assessed - it is graded on its stated system, not on its execution, which is not the same standard applied to the other four. source 2026
three of four studies give LIN28B a non-pro-leukaemic sign; the entry corrects a title-only counter-evidence citation
the sign dossier is the evidence base the lineage hypothesis tries to reconcile
most relevant system, least readable; the design-quality gradient runs against it
the mechanism recovered by reading the counter-evidence full text
most human paediatric AML including MLLr AML do not express LIN28B and are let-7-high
the let-7a route is the mechanism offered for the oncogene reading
liability 2 - the sign dossier