IKZF3 / Aiolos (Q9UKT9) - the B-lymphoid transcription factor that goes UP under both inhibitors, regulates BCL2, and is a degradable drug target

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.

• A menin inhibitor RAISES IKZF3, and IKZF3 activates BCL2 through mapped promoter sites. If that transmits, the drug is raising the survival protein that its most likely partner agent - venetoclax - is meant to block. The opposite reading is equally available: raising BCL2 could deepen BCL2 dependence and increase venetoclax sensitivity. The data here cannot choose, and one qPCR would.
• IKZF3 is a cereblon neosubstrate, so it is a DEGRADABLE target - but degrading it removes a B-cell differentiation factor, and differentiation is the mechanism by which menin inhibitors are thought to work. Combining a menin inhibitor with an IMiD or CELMoD could therefore remove the very output the menin inhibitor is producing.
• The Ikaros-family dependency reported in MLL-rearranged leukaemia is an IKZF1 dependency, and IKZF3 is not a dependency in 1,253 DepMap screens. Anyone building an IKZF3-directed strategy from the PROTAC paper would be reading a claim about its paralogue.
• The whole IKZF3 rise is from a baseline of 1.58 log2 CPM. A gene rising 2.5-fold from near-nothing may still be functionally absent, and the graph should not treat +1.34 on a barely detected transcript the same way it treats CHAC1's -5.11 on a well-expressed one.

1. Does BCL2 mRNA or protein rise in RS4;11 under VTP-50469? Aiolos activates BCL2 through promoter sites and IKZF3 rises under the drug, and venetoclax plus a menin inhibitor is an active clinical combination - so this is one lane on a blot with a real combination hanging on it, and BCL2 was not in this beat's pre-registered panel.
2. WHICH IKZF3 isoform rises? Isoform 12 is cytoplasmic and cannot act as a transcription factor while isoforms 1, 3 and 11 are nuclear, and both platforms used here are isoform-blind. Isoform-resolved qPCR or long-read sequencing on drug-treated RS4;11 would settle it, and the entire BCL2 argument depends on the answer.
3. Is IKZF3 a dependency in KMT2A-rearranged B-ALL lines specifically? The pan-leukaemia DepMap mean (-0.098) pools myeloid and lymphoid lines, and a B-lineage-confined dependency in a handful of lines would be invisible in it.
4. Does an IMiD or CELMoD combine with or oppose a menin inhibitor in KMT2A-rearranged B-ALL? A menin inhibitor RAISES the very protein lenalidomide-class drugs degrade, which is either a rationale for the combination or evidence that the two agents fight each other on this axis.

1. E1established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.dataset

   IKZF3 rises 1.08 log2 on its best-expressed GPL570 probe under MI-2-2 in RS4;11 (p 0.011, DMSO 6.31) and 2.73 in SEM (p 0.008), while KMT2A-wild-type REH moves -0.32; under VTP-50469 in RS4;11 at day 7 it rises 1.34 from a baseline of only 1.58 log2 CPM, with EPZ5676 giving +1.46; and in myeloid MOLM13 it moves the other way at -0.41.

   Q106 pre-registered ISR panel run over the cached GSE63664 (MI-2-2 / EPZ004777, SEM / RS4;11 / REH, GPL570 RMA) and GSE127506+GSE127508 (VTP-50469 330 nM / EPZ5676 1 uM, RS4;11 and MOLM13, d2 and d7, 3-prime-end RNA-seq) matrices. Script analyses/does-atf4s-own-transcript-move-under-a-menin-inhibitor/run.py, results.json, panel frozen in panel-preregistration.json BEFORE any expression value was read. IKZF3 probes 227030_at, 236301_at, 221092_at; RNA-seq row IKZF3, both lines, both days. · NCBI GEO, re-analysed here 2026
   ▸What this evidence cannot show

   LOW BASELINE on both platforms: the RS4;11 RNA-seq DMSO level is 1.58 log2 CPM, barely above the 1.0 detection floor used here, and three of four array probes are at background (2.99-5.25 log2 RMA). The defensible claim is a rise from a low baseline, NOT that IKZF3 becomes abundant. REH's baseline (4.43) is also at background so its -0.32 is a weak negative control. ISOFORM-BLIND: both platforms sum across IKZF3's many splice variants, which differ in compartment, so a rise in total transcript does not establish a rise in the nuclear, transcription-factor-competent form. RS4;11 array DMSO arm n=2.

   ▸Evidence that argues against this (1)

   • EPZ5676 gives +1.46 against VTP-50469's +1.34, so like most of this beat's panel the IKZF3 rise is shared with DOT1L inhibition and is not a menin-interface readout. The most parsimonious account of all of it - including the MOLM13 reversal - is that blocking the fusion's transcriptional output by any route lets a lineage-appropriate differentiation programme re-emerge, which is the drug class's established mechanism and requires no gene-specific explanation. NCBI GEO, re-analysed here 2026
2. E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

   Aiolos interacts with Ras in IL-2-stimulated and IL-2-deprived cells; IL-2 controls its cellular distribution and induces the tyrosine phosphorylation required for dissociation from Ras; functional Aiolos-binding sites in the BCL2 promoter activate a luciferase reporter and mutating them inhibits transactivation; and co-transfection shows Aiolos induces Bcl-2 expression and prevents apoptosis in IL-2-deprived cells.

   Aiolos transcription factor controls cell death in T cells by regulating Bcl-2 expression and its cellular localization. · Romero F, Martinez-A C, Camonis J, et al. · EMBO Journal 1999
   PMID 10369681DOI
   ▸What this evidence cannot show

   T CELLS, IL-2 deprivation, transfection and promoter-reporter assays, 1999 - and Aiolos is a B-LYMPHOID factor, so the BCL2 result is in the wrong lineage for this project. Reporter transactivation with site mutagenesis is good evidence of a functional site but is not an endogenous-locus experiment, and there is no ChIP. The Ras interaction half of the paper is not used here. Abstract read; figures not.

   ▸Evidence that argues against this (1)

   • If a menin inhibitor raises IKZF3 and IKZF3 activates BCL2, the effect on a venetoclax combination could go either way - raising BCL2 could oppose the partner drug, or could raise BCL2 dependence and increase sensitivity. The data here choose neither, because BCL2 was not in the pre-registered panel and no venetoclax arm exists in either dataset. The argument is a hypothesis with a trivial experiment attached, not a finding. NCBI GEO, re-analysed here 2026
3. E3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

   IKZF1 and IKZF3 are the well-established cereblon neosubstrates of immunomodulatory drugs, and a cereblon-recruiting dual WDR5/Ikaros PROTAC (MS40) degrades WDR5 together with IKZF1 and IKZF3, causing dissociation of the MLL/KMT2A complex from chromatin with decreased H3K4me2; in MLL-rearranged leukaemias, described in that work as exhibiting high IKZF1 expression and dependency, co-suppression of WDR5 and Ikaros is superior to a WDR5 protein-protein-interaction inhibitor, to non-PROTAC analogue controls, and to a matched VHL-based WDR5 PROTAC that degrades WDR5 but not Ikaros, and MS40 suppresses primary patient leukaemia cells in vitro and patient-derived xenografts in vivo.

   Discovery of a dual WDR5 and Ikaros PROTAC degrader as an anti-cancer therapeutic. · Li D, Yu X, Kottur J, Gong W, et al. · Oncogene 2022
   PMID 35525905DOI
   ▸What this evidence cannot show

   CAREFUL READING REQUIRED, AND THIS IS THE CLAIM MOST AT RISK OF BEING MISQUOTED: the expression-and-dependency statement in that abstract is about IKZF1, NOT IKZF3. IKZF3 is degraded alongside IKZF1 because both are CRBN neosubstrates, and the paper's superiority argument is for co-suppression of WDR5 and 'Ikaros' collectively. NOTHING here establishes an IKZF3-specific dependency in MLL-rearranged leukaemia, and this beat's own DepMap query finds IKZF3 is not a dependency (-0.1016 mean, 3.1% of lines below -0.5). 'MLL-rearranged' is not resolved into fusion partners. Abstract read; figures not. Already cited in the graph: checked, and this is a new use.

   ▸Evidence that argues against this (1)

   • The PROTAC's effect could be entirely a WDR5 effect plus an IKZF1 effect, with IKZF3 degradation a bystander - and the matched VHL PROTAC control that degrades WDR5 but not Ikaros does not separate IKZF1 from IKZF3. So the strongest available evidence that Ikaros-family degradation matters in MLL-rearranged leukaemia does not implicate IKZF3 specifically, and this beat's DepMap figure argues against it. DepMap / Open Targets Platform 2026
4. E4strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

   UniProt Q9UKT9 is the single reviewed human IKZF3 record, 509 aa, synonym ZNFN1A3, recommended name 'Zinc finger protein Aiolos', with ISOFORM-RESOLVED subcellular localisation: isoforms 1, 3 and 11 nuclear, isoform 14 nuclear and cytoplasmic, and ISOFORM 12 CYTOPLASMIC ONLY, all ECO:0000269 from PubMed 17646674 - a paper whose stated purpose is to investigate directly the cellular localisation of Aiolos isoforms, their heterodimerisation with Ikaros, their association with HDAC-containing complexes and the resulting effects on histone modification, and which reports that full-length Ikaros-family isoforms localise to heterochromatin.

   Combinatorial effects of splice variants modulate function of Aiolos. · Caballero R, Setien F, Lopez-Serra L, et al. · Journal of Cell Science 2007
   PMID 17646674DOI
   ▸What this evidence cannot show

   Resolved live 2026-09-22: exactly one reviewed record (rule 62 satisfied), re-fetched alone with name, length and primary symbol compared back: PASS. RULE 91: this is the GOOD case - the coded citation is a dedicated localisation study of this protein's isoforms, unlike most of this beat's panel where the code points at a paper about something else. CONSEQUENCE FOR THIS BEAT: 'IKZF3 is nuclear' is not an available statement, and both the array and the 3-prime-end RNA-seq are isoform-blind, so the measured rise cannot be assigned to a compartment. Abstract read; the imaging figures were not.

   ▸Evidence that argues against this (1)

   • Isoform-resolved localisation from one 2007 paper using transfected constructs of novel splice variants may not reflect the isoform distribution of endogenous IKZF3 in a leukaemia cell, and the number of Aiolos isoforms reported has grown since. The practical warning - that a total-transcript measurement cannot be assigned to a compartment - stands regardless of how accurate the individual isoform calls are. UniProt Knowledgebase 2026
5. E5established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database

   In DepMap 26.06, IKZF3 has a mean CRISPR gene effect of -0.1016 across 1,253 screens with 3.1% of lines below -0.5 and a leukaemia mean of -0.0983 across 57 leukaemia screens.

   DepMap CRISPR (Chronos) gene effect via the Open Targets Platform GraphQL depMapEssentiality field; dataVersion year 26 month 06, read back from meta{dataVersion} in the same session · DepMap / Open Targets Platform 2026
   ▸What this evidence cannot show

   Release recorded per rule 83; IKZF3's screen count is 1,253, gene-specific. Leukaemia screens by substring match with ids persisted per rule 90. This figure directly qualifies E3: the Ikaros-family dependency described in MLL-rearranged leukaemia is an IKZF1 dependency, and IKZF3 itself is not a dependency in the 57 leukaemia lines screened. The eleven KMT2A-rearranged lines were not separated out.

   ▸Evidence that argues against this (1)

   • A pan-leukaemia mean pools myeloid and lymphoid lines, and IKZF3 is a B-lymphoid factor whose dependency, if any, would be confined to B-lineage lines - possibly to a handful of them. A flat mean across 57 mixed leukaemia screens cannot exclude a dependency in B-ALL, and this beat did not stratify by lineage or by KMT2A status. DepMap / Open Targets Platform 2026
6. E6established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database

   PubMed returns 2 records for IKZF3 jointly with an eight-operand KMT2A-fusion list - PMID 35525905 (the dual WDR5/Ikaros PROTAC) and PMID 38139431 (copy-number alterations in adult Ph-negative B-ALL) - and 0 for IKZF3 with a thirteen-name menin-inhibitor list with a non-empty querytranslation and 'No items found.'; the alias Aiolos reaches 260 records of which 85 also mention IKZF3 (Jaccard 0.1683) and ZNFN1A3 reaches 6, and fusion-scoped Aiolos returns a valid zero while fusion-scoped IKZF3 returns the two.

   NCBI PubMed esearch, run 2026-09-22, with querytranslation and warninglist read back from the response for every query quoted · NCBI PubMed 2026
   ▸What this evidence cannot show

   Both on-topic titles were fetched and read and the PROTAC abstract was obtained. Rule 3 satisfied on the zeros. NOTE THE DIRECTION: here the MODERN symbol is the better instrument and the historical alias returns nothing fusion-scoped - the exact opposite of the PPP1R15A/GADD34 case in this same beat, which is why 'always widen the name list' is not a usable rule. Rule 94 form; rule 95 observed; rule 80: titles and abstracts only.

   ▸Evidence that argues against this (1)

   • IKZF3's clinical importance is overwhelmingly in multiple myeloma, where it is the lenalidomide neosubstrate, and that literature is invisible to a KMT2A-scoped query. The two fusion-scoped records understate what is known about degrading this protein in patients; the absence is about the KMT2A field, not about the drug target. Li D et al. 2022

• regulates →BCL2 — the one co-dependency in this beat that already has an approved drug pointed at it

  functional Aiolos-binding sites in the BCL2 promoter activate a reporter, mutating them blocks transactivation, and Aiolos induces Bcl-2 and prevents apoptosis in IL-2-deprived T cells (PMID 10369681) - wrong lineage, trivial experiment to repeat

• related to →KMT2A::AFF1 (MLL-AF4), t(4;11)(q21;q23)

  rises in BOTH t(4;11) lines under both compounds (SEM +2.73, RS4;11 +1.08 array / +1.34 RNA-seq) and FALLS in myeloid MOLM13 (-0.41) - lineage-coupled, from a low baseline

• ← related toVenetoclax in KMT2A-rearranged leukemia — the apoptotic partner every other drug needs

  UNTESTED COMBINATION RISK: a menin inhibitor raises IKZF3, IKZF3 activates BCL2, and venetoclax plus a menin inhibitor is an active clinical combination. No BCL2 measurement exists under a menin inhibitor in these datasets

• related to →WDR5 — the partner that is not a partner: it binds the half of KMT2A the fusion throws away

  a cereblon-recruiting dual WDR5/Ikaros PROTAC degrades WDR5 with IKZF1 and IKZF3, dissociating the MLL/KMT2A complex from chromatin with decreased H3K4me2, and is superior to WDR5 degradation alone in MLL-rearranged leukaemia (PMID 35525905)

• related to →Pro-B cell — the stage at which KMT2A::AFF1 leukaemia is arrested, and the phenotype the diagnosis is made on

  IKZF3/Aiolos is essential for B-cell differentiation, proliferation and maturation to an effector state; its drug-induced rise is confined to the two B-lineage lines

• related to →The differentiation block — what it actually is, operationally, in KMT2A::AFF1 leukaemia

  the rise is lineage-coupled (up in both B-lineage t(4;11) lines, down in myeloid MOLM13), which is what re-emergence of a lineage-appropriate differentiation programme looks like