CDK4 - the other half of the class name, and it is NOT the gene this disease needs

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.

• This entry is an argument against half of a drug for which no half exists. Acting on it requires CDK6-selective chemical matter that is not in the clinic, so the operational use is interpretive - it says how to read a dual-agent result, not what to prescribe.
• Paralogue buffering cuts both ways. If CDK6 is inhibited, CDK4 may take over; a CDK6-selective agent would then be escaped by a kinase the dual agent happens to cover. Removing the CDK4 arm could remove a benefit nobody attributed to it.
• CDK4 is more broadly essential across 1,258 lines than CDK6 [E2]. Any argument that CDK4 inhibition is harmless because this disease does not need it is a lineage-scoped argument, and a systemic drug is not lineage-scoped.
• A negative knockdown result (E3, E4) is weaker evidence than a positive one, and both published CDK4 negatives in this disease are knockdowns. The DepMap number is a knockout and agrees, which is the strongest form this argument currently takes - but all three are genetic and none is pharmacological.

1. Does CDK4 become required in KMT2A-rearranged cells after CDK6 is inhibited? Paralogue buffering predicts it should, and if so the dual agent is right for the wrong reason and a CDK6-selective agent would fail by escape rather than by potency. A CDK4 knockout in a CDK6-inhibited KMT2A::AFF1 line answers it directly.
2. Is there any KMT2A-rearranged context in which CDK4 rather than CDK6 carries the cyclin D partner? CCND3 is the leukaemia cyclin and its kinase partner in these cells has not been assigned by co-IP in any source read this cycle.
3. Would a CDK4-sparing agent actually reduce toxicity, or is the dose-limiting neutropenia entirely CDK6-mediated? If it is, CDK6-selectivity buys specificity of rationale and nothing in safety.

1. E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.dataset

   In a genome-wide scan of 17,728 protein-coding genes for selective requirement in 11 Cellosaurus-annotated KMT2A-rearranged leukaemia lines versus 43 other leukaemia lines, CDK4 ranks 16,052nd with a difference of +0.0716 (KMT2A-r mean -0.3742, other leukaemia mean -0.4458), while CDK6 ranks 13th with a difference of -0.5266.

   Genome-wide KMT2A-rearranged selectivity scan, 17,728 genes (DepMap Chronos via Open Targets); analyses/depmap-kmt2a-selective-scan-full/results.json · Broad Institute DepMap; Cellosaurus (SIB); Open Targets Platform; analysis run by this project · Databases and derived analysis 2026
   ▸What this evidence cannot show

   This project's own unreviewed computation, re-read from the persisted results file on 2026-09-20, not re-run. Eleven index lines, none infant-derived; the index arm is myeloid-enriched and the comparison is not lineage-conditioned. A rank near the bottom of a 17,728-gene list is a weak measurement in absolute terms - the difference is small and the interesting fact is its SIGN relative to CDK6, not its magnitude.

   ▸Evidence that argues against this (1)

   • CDK4 and CDK6 are reported to have redundant functions in cell-cycle progression, so a single-gene CDK4 knockout in cells that still express CDK6 understates CDK4 and the +0.0716 cannot be read as CDK4 being dispensable. Scheicher R et al. 2015
2. E2established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database

   Fetched live on 2026-09-20, DepMap Chronos gene effect for CDK4 across 1,258 cell lines has mean -0.7371 and median -0.5645, with 58.4% of lines at or below -0.5; restricted to 57 leukaemia-annotated lines the mean is -0.4464. For CDK6 the same query gives -0.6321 across all lines and -1.3767 across leukaemia lines. CDK4 is therefore more broadly required across tissues and markedly less required in haematopoietic lines than CDK6.

   DepMap CRISPR (Chronos) gene effect for CDK4 (ENSG00000135446) and CDK6 (ENSG00000105810), retrieved via the Open Targets Platform GraphQL API, depMapEssentiality.screens · Broad Institute DepMap; Open Targets Platform; query run by this project 2026-09-20 · Database query 2026
   ▸What this evidence cannot show

   The leukaemia subset is a coarse text match on the screen's own diseaseFromSource field (leukemia/leukaemia/lymphoblastic/myeloid), not a curated lineage annotation, and is not conditioned on fusion status. DepMap release version not stated by the API. Knockout fitness in immortalised lines over roughly 21 days; no infant-derived line in the panel. A database fact opposable on scope.

   ▸Evidence that argues against this (1)

   • Every approved agent in this class engages both kinases, so a CDK4-versus-CDK6 gene-effect difference does not describe what palbociclib, ribociclib or abemaciclib do in a patient, and cannot be used to predict a clinical result. Raetz EA et al. 2023
3. E3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

   In MLL-rearranged infant ALL, knockdown of CDK6 inhibits growth of the KMT2A::AFF1-positive line SEM while CDK4 knockdown has virtually no effect.

   MLL fusion-driven activation of CDK6 potentiates proliferation in MLL-rearranged infant ALL · van der Linden MH, Willekes M, van Roon E, Seslija L, Schneider P, Pieters R, Stam RW · Cell Cycle 13:834-844 2014
   PMID 24736461DOI
   ▸What this evidence cannot show

   Abstract read this cycle. One cell line for the CDK4-versus-CDK6 comparison, and SEM is a t(4;11) line from a five-year-old (CVCL_0095) rather than infant-derived despite the paper's infant-ALL framing. A knockdown is partial, so a negative CDK4 arm is weaker evidence than a positive CDK6 arm.

   ▸Evidence that argues against this (1)

   • A knockdown with no phenotype can reflect incomplete depletion rather than dispensability, and no CDK4 protein-level depletion efficiency is reported in the abstract. van der Linden MH et al. 2014
4. E4strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

   In murine and human MLL-AF9-driven AML, cells are exceptionally reliant on CDK6 but not CDK4.

   Requirement for CDK6 in MLL-rearranged acute myeloid leukemia · Placke T, Faber K, Nonami A, Putwain SL, Salih HR, Heidel FH, et al. · Blood 124:13-23 2014
   PMID 24764564DOI
   ▸What this evidence cannot show

   Abstract read this cycle. Murine and human MYELOID MLL-AF9 disease, not infant B-lineage KMT2A::AFF1. Rule 73: the system is named in the claim because it is not the index disease.

   ▸Evidence that argues against this (1)

   • This is a second knockdown-based negative for CDK4 from the same era, so the two independent confirmations share a method and its limitation rather than being methodologically independent. van der Linden MH et al. 2014
5. E5established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper

   CDK6 and CDK4 have redundant functions in regulating cell-cycle progression, which is the stated premise of the work that then identifies a CDK6-specific, cell-cycle-independent role in haematopoietic and leukaemic stem cells.

   CDK6 as a key regulator of hematopoietic and leukemic stem cell activation · Scheicher R, Hoelbl-Kovacic A, Bellutti F, Tigan AS, Prchal-Murphy M, Heller G, et al. · Blood 125:90-101 2015
   PMID 25342715DOI
   ▸What this evidence cannot show

   Abstract read this cycle. The redundancy statement is the paper's framing sentence rather than a measurement made in it; it is cited here as the field's standing position, which is why it is graded established for the general claim and no higher.

   ▸Evidence that argues against this (1)

   • If the two kinases were fully redundant, the 0.93 gene-effect-unit separation between them in 57 leukaemia lines could not exist. The redundancy is therefore partial and tissue-dependent, which is the position this entry takes. Broad Institute DepMap; Open Targets Platform 2026
6. E_q95_cdk4_loc1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

   CDK4's compartment is CELL-CYCLE-PHASE DEPENDENT and the itinerary is resolved to the nuclear pore: immunolocalisation shows cyclin D2 and Cdk4 forming a complex IN THE CYTOPLASM and approaching the nucleus, accumulating on the CYTOSOLIC SURFACES OF THE NUCLEAR PORES, then being ARRESTED AT THE NUCLEAR MEMBRANE before the nucleus reaches a critical size, and finally released into the nucleus where the complex co-localises with pRb, leading to pRb phosphorylation and DNA synthesis; the translocalisation depends on the G1-S transition, and a truncated cyclin D2 unable to associate fully with Cdk4 loses the ability to be released into the nucleus.

   Migratory localization of cyclin D2-Cdk4 complex suggests a spatial regulation of the G1-S transition. · Wang Z, Xie Y, Zhang L, Zhang H, An X, Wang T, Meng A · Cell Struct Funct 2008
   PMID 18827403DOI
   ▸What this evidence cannot show

   The complex studied is cyclin D2-CDK4 under OVEREXPRESSION of both partners, in mammalian cell lines not named in the abstract, and cyclin D1 and D3 are not tested - so the itinerary may be D-cyclin-specific. It is nonetheless a dedicated localisation study with the phase dependence as its subject, which makes it one of the better-sourced compartment citations in this tranche, and the partner-dependence of nuclear entry makes CDK4 the FOURTH complex-dependent nuclear residence in this localisation series after SGF29, Paf1C/CTR9 and TRRAP.

   ▸Evidence that argues against this (1)

   • A SINGLE RESTING-STATE ANSWER FOR THIS GENE WOULD BE WRONG IN BOTH DIRECTIONS, AND THE DATABASE ITSELF SAYS SO - WHICH IS WHY THE PHASE HAS TO BE IN THE CLAIM. UniProt's note reads 'Cytoplasmic when non-complexed', so an unpartnered CDK4 is a cytoplasmic protein, and the nuclear pool exists only for the assembled cyclin D-CDK4 complex past the G1-S transition. Two consequences this graph should carry. First, any inference from a CDK4 immunofluorescence image without cell-cycle staging is uninterpretable. Second, and more useful: because nuclear entry requires the cyclin, the compartment is a readout of D-cyclin availability - and the one MLL-AF9 result in this space reports elevated CDK4 AND CDK6 protein in GFI1-36N leukaemic cells with greater palbociclib susceptibility, measured by proteomics, immunofluorescence and immunoblotting, with no compartment or staging reported and no separation of the two kinases. Vorwerk J et al. 2022
7. E_q95_cdk4_loc2weak confidence 2 of 5: Preliminary, small, indirect, or not yet peer reviewed.database

   UniProt's subnuclear clause for CDK4 - 'Also present in nucleoli and heterochromatin lumps' - is UNVERIFIABLE by this beat and UNCORROBORATED in the reachable literature. The phrase is attributed to ECO:0000269 PubMed 18827403, whose abstract does not contain it; the full text was pursued through the standing five-route test and could not be obtained as text (Europe PMC returns no PMCID, and the DOI resolves to a J-Stage page serving 3,357 characters of abstract-level HTML while the PDF's text layer could not be extracted). Independently, CDK4[tiab] against nucleolus or nucleolar returns 29 records and the title-checked first twenty contain no CDK4 compartment measurement at all.

   NCBI E-utilities esearch db=pubmed; every count stored with the querytranslation read back, in data.localization_bounded_absence · NCBI E-utilities · Databases 2026
   ▸What this evidence cannot show

   This does NOT claim the paper contains no nucleolar observation - the same discipline tranche 5 applied to CTR9's unreadable speckle source applies here. What is established is that no reachable text supports the phrase and that the independent literature bound is empty, so the graph should not repeat 'CDK4 is nucleolar' as though it were measured. Status of this item is deliberately weak rather than strong for that reason.

   ▸Evidence that argues against this (1)

   • AND THE LITERATURE BOUND IS ITSELF CONTAMINATED IN THE FALSE-POSITIVE DIRECTION, WHICH IS WHY THE TITLE CHECK MATTERED MORE THAN THE COUNT. Of the 29 records, the title-checked first twenty are SNHG1, SNHG4, SNHG10, SNHG12, SNHG17, SNORD9 and nucleolar-stress-signalling papers - the 'nucleolar' token matching 'small NUCLEOLAR RNA host gene' nomenclature exactly as symbol-collisions row 105 predicts - plus ribosomal-protein and NF-kB-nucleolus reviews. The single record that is genuinely about a CDK4 and a nucleolus is PubMed 40383750, in which C. elegans CDK-4 REGULATES nucleolar SIZE at the cost of late-life fitness: a regulatory relationship in a nematode, not a human compartment. So an unscreened count of 29 would have read as a modest supporting literature for the UniProt phrase, and there is none. Fourth gene in this tranche where that token misfired, alongside USP8 (2 of 2 off topic), YAP1 (9 of 16) and SREBF1. Webster R et al. 2025
8. E_q95_cdk4_loc3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.database

   Bounded 2026-09-22: no CDK4 compartment has been measured in any KMT2A-fusion cell. The gene clause over CDK4 and 'PSK-J3' returns 13,418 records and intersects the nine-term KMT2A/MLL fusion clause at 25; with a localisation clause the gene clause returns 1,388 and with the fusion clause added 2, which title-check to 0 on topic - one is a gastric-cancer methylation review and the other is the GFI1-36N/palbociclib paper, which reports protein levels rather than compartments; with a subnuclear clause the gene clause returns 52 and with the fusion clause added 0.

   NCBI E-utilities esearch db=pubmed; every count stored with the querytranslation read back, in data.localization_bounded_absence · NCBI E-utilities · Databases 2026
   ▸What this evidence cannot show

   Titles and abstracts only; supporting figures not searched (rule 80), and for a cell-cycle kinase whose nuclear entry is staged this is a real limitation - a staged immunofluorescence panel in a leukaemia paper would be a supporting figure. The gene clause is large and includes CDK4 as a pharmacological target rather than a subject, so the zeros sit behind a wider net than labelled.

   ▸Evidence that argues against this (1)

   • THE ABSENCE IS CORRECT AND THIS ENTRY'S OWN VERDICT MAKES IT NEARLY MOOT, WHICH IS THE HONEST THING TO SAY RATHER THAN LISTING IT AS A GAP. This entry exists to record that CDK4 is NOT the gene the KMT2A-r rationale rests on: it is more broadly essential across tissues than CDK6, markedly less required in leukaemia lines (leukaemia mean Chronos -0.446 against an all-lines -0.737), and its KMT2A-r selectivity difference is POSITIVE at rank 16,052 of 17,728 - so the CDK4 arm of a dual inhibitor is a toxicity contribution without a matching selectivity contribution. A staged CDK4 localisation experiment in a KMT2A-rearranged cell would be a well-designed measurement of a gene this disease does not appear to need. The compartment result that would matter is CDK6's, and this graph holds cell-state-cell-cycle-g1-cdk6 for that arm. Vorwerk J et al. 2022

• ← related toCDK6 - the kinase the KMT2A-r rationale actually rests on, and the pan-essentiality flag and the selectivity are both true

  paralogous G1 kinases at opposite ends of the KMT2A-r selectivity ranking (13 vs 16,052 of 17,728)

• treated by →CDK4/6 inhibition in KMT2A-rearranged leukemia — the class was already tried, and the trial never reported

  engaged by every approved agent in the class but not preferentially required in this disease

• ← inhibitsCDKN2A - one gene, two reviewed UniProt records with the same primary symbol, and a deletion that is three times rarer here than in the infant leukaemias next door

  p16INK4a is named 'cyclin-dependent kinase 4 inhibitor A' on its own UniProt record

• ← related toBMI1, CDKN2A and CDK6 are one circuit, not three genes - and in KMT2A-rearranged disease it has a loop nobody has measured

  the other p16INK4a target; included so the circuit is not written as CDK4/6

• ← supportsLive DepMap on fourteen genes: CDK6's pan-essentiality is a lineage artefact of pooling, and not one gene of the homeodomain cofactor code is a dependency anywhere

  CDK4 is the broader dependency (-0.7371) and the shallower one in leukaemia (-0.4464)

• related to →MLL was reported at regions attached to the nuclear envelope in 2000, which is indirect evidence at the lamina that this graph currently records as absent

  CDK4's compartment is a PHASE, resolved to the pore: cytoplasmic when non-complexed, the cyclin D2-CDK4 complex accumulates on the CYTOSOLIC SURFACES OF NUCLEAR PORES, is ARRESTED AT THE NUCLEAR MEMBRANE, and is released into the nucleus at the G1-S transition to meet pRb (PMID 18827403). A truncated cyclin D2 that cannot bind CDK4 abolishes release - fourth complex-dependent nuclear residence in this series (Q95 2026-09-22)

• related to →The G1 state in KMT2A-rearranged leukaemia — where CDK6 does a job that is not about the cell cycle at all

  The compartment question that matters is CDK6's, not CDK4's. CDK4 is LESS required in leukaemia lines than elsewhere (Chronos -0.446 against -0.737) with a POSITIVE KMT2A-r selectivity difference at rank 16,052/17,728, and the one MLL-AF9 result uses palbociclib, which is pan-CDK4/6 and moves both kinases together (PMID 36003783). A staged CDK4 imaging experiment would measure a gene this disease does not need (Q95 2026-09-22)

• related to →The nucleolus — the one compartment with direct published images of a KMT2A fusion inside it

  A NEGATIVE entry for this node. UniProt's 'Also present in nucleoli and heterochromatin lumps' is attributed to PMID 18827403, is absent from its abstract, and is unverifiable through all five routes (no PMCID; J-Stage serves 3,357 characters of abstract HTML; PDF text layer will not extract). The independent bound is empty: 29 records, 0 human CDK4 compartment measurements, contamination by SNHG/SNORD nomenclature. Do not repeat it as measured (Q95 2026-09-22)