Oncofusion

start / graph / gene-csnk2a1

Genemoderate confidence 3 of 5: One strong study, or several smaller consistent ones.Q22-phospho-contacts

CSNK2A1 (casein kinase 2 alpha) - the one kinase that turns up three separate times in KMT2A biology

also called CK2 · CKII · CK2alpha · casein kinase II subunit alpha · CSNK2A1

CK2 is a constantly-active enzyme that sticks phosphate groups onto other proteins. In KMT2A leukaemia it shows up three separate times: it tightens the fusion's grip on a partner, it controls how the healthy copy of KMT2A is cut in half, and it switches on a protein that leukaemia stem cells need.

The same thing, for a scientist

CSNK2A1 (P68400, CSK21_HUMAN, 391 aa) is the catalytic alpha subunit of the constitutively active Ser/Thr kinase CK2. It reaches KMT2A biology through **three mechanistically independent routes**: (1) it phosphorylates the KMT2A IBD-binding motif at **S136/S142**, raising LEDGF/p75 affinity, and the phospho-null fusion is transformation-impaired; (2) it phosphorylates KMT2A at **T2724/S2726**, immediately C-terminal to taspase1 cleavage site CS2, and CK2 inhibition blocks cleavage, stabilises full-length MLL1 and **displaces MLL chimeras from chromatin**, delaying leukaemia in an MLL-AF9 mouse; (3) in a 2026 Blood study it phosphorylates **DEK at S301/303/306/307**, and the clinical-stage CK2 inhibitor CX-4945 selectively depletes leukaemia stem cells in MLL-AF9 models while sparing normal HSPCs. Route 1 acts on the **retained** fragment, route 2 on the **lost** fragment (and therefore on the wild-type allele only), route 3 not on KMT2A at all. **This is the only kinase in the beat that touches the fusion, the wild-type allele and the stem-cell program by three different mechanisms, and it has a drug in the clinic.**

Go deeper
Three routes, and why they must not be conflated
routesubstratesitefate in fusionconsequencesource
1KMT2AS136, S142retainedraises LEDGF IBD affinity; phospho-null fusion loses transformation capacity1
2KMT2AT2724, S2726lostlicenses taspase1 cleavage; inhibition stabilises full-length MLL1 and evicts the chimera2
3DEKS301/303/306/307n/asustains the DEK-GABPA-PBX3 stem-cell program in MLL-AF93

Routes 1 and 2 predict opposite responses to CK2 inhibition at the level of the fusion's chromatin grip - route 1 says inhibition loosens the fusion's own LEDGF contact, route 2 says inhibition works by strengthening a competitor. Both end at 'CK2 inhibition is good in KMT2A-rearranged leukaemia', and that agreement is exactly what makes them easy to confuse. Any experiment reading a CX-4945 phenotype has to say which route it is measuring.

What is known about the drug side

CX-4945 (silmitasertib) is the clinical-stage CK2 inhibitor used in both the taspase1 study and the 2026 DEK study. In the taspase1 work it inhibited CKII phosphorylation of the MLL wild-type peptide with an IC50 of 0.011 uM in vitro, increased full-length MLL1 without altering MLL-AF4 protein levels in SEM cells, reduced genome-wide MLL-AFF1, AFF4 and ELL2 occupancy, and extended survival in an MLL-AF9 syngeneic mouse by oral gavage and by intraperitoneal injection 2. In the DEK work, CX-4945 depleted leukaemia stem cells while sparing normal HSPCs and synergised with venetoclax 3.

That is an unusually complete preclinical package for a target this graph has not previously carried as a node.

Two caveats: the tools cannot tell the paralogues apart, and the symbols cannot either

The chemistry does not attribute. The CK2 routes recorded in this entry were established with CX-4945, TBB and DMAT, all of which inhibit the CK2 catalytic subunits as a class. Attributing those results to CSNK2A1 rather than to its paralogue CSNK2A2 is a naming convention, not a measurement. Any claim here that turns on which alpha subunit is responsible needs a genetic experiment that has not been done.

And the nomenclature is actively dangerous. Fetched from HGNC on 2026-09-20:

recordsymbolalias symbolsUniProtlength
HGNC:2457CSNK2A1Cka1, Cka2P68400391 aa
HGNC:2459CSNK2A2CSNK2A1, CK2alpha'P19784350 aa

CSNK2A1 is an alias symbol of CSNK2A2. A lookup that matches alias fields - which most do by default - can resolve the string CSNK2A1 to the other paralogue, returning a 350-residue protein where a 391-residue one was meant. The two are the two catalytic subunits of the same kinase, so nothing about the returned record looks wrong.

This is now rule 63 in the queen brief, and it is a sharper form of the trap already logged there: it is not a loose substring match but a curated alias that points at a different gene of the same family.

A fourth route, and it is a histone one: CK2 writes H4 phospho-serine-1, and licenses MLL1 and SEC recruitment through LEDGF

This entry describes three mechanistically independent routes from CK2 into KMT2A biology and contains the word histone zero times. There is a fourth route and it is the epigenetic one, and it has two halves - a mark CK2 actually writes, and a recruitment step in this project's own spine.

CK2 is a histone kinase. In a study of NuA4 regulation, casein kinase 2 phosphorylates histone H4, and the consequence is a phospho/acetyl switch on the H4 tail: phosphorylation of serine 1 (S1P) strongly decreases the ability of the NuA4 H4/H2A acetyltransferase complex to acetylate H4 peptides, while arginine-3 dimethylation combined with S1P alleviates that repression. Chromatin from cells treated with DNA-damage-inducing agents shows increased S1P with a concomitant decrease in H4 acetylation; CK2 associates with the Rpd3 deacetylase complex, giving a physical connection between S1 phosphorylation and unacetylated H4 tails; and ChIP links local H4 phosphorylation to its deacetylation during both transcription and DNA repair, with time-course data placing H4 phosphorylation after NuA4 action during double-strand-break repair, at the step of chromatin restoration E8. This is a genuine writer activity and, of the fourteen entries in this tranche, the only one. It is also yeast - NuA4 and Rpd3 are S. cerevisiae complexes - and that limit is load-bearing.

And a second, non-histone substrate with a mark-level readout. CK2 interacts with and phosphorylates the major histone chaperone Spt6, and that phosphorylation is required for Spt6's cellular levels, for suppression of histone H3 turnover, and for inhibition of spurious transcription. CK2 depletion causes a replication-independent increase in H3K56 acetylation and global activation of H3 turnover in coding regions, and CK2 inhibits global cryptic promoters driving both sense and antisense transcription E9. Also yeast, and again a chromatin consequence reached through a non-histone substrate - the same shape this tranche recorded for CDK9 and H2Bub.

The on-beat half: CK2 licenses MLL1 and SEC onto chromatin through LEDGF. At the HIV provirus, LEDGF/p75 suppresses transcription during latency by recruiting the PAF1 complex and pausing Pol II. On latency reversal, the MLL1 complex competitively displaces PAF1 through a casein-kinase-II-dependent association with LEDGF/p75, and depleting or pharmacologically inhibiting CKII prevents PAF1 dissociation and abrogates the recruitment of both MLL1 and the Super Elongation Complex E10. Every object in that sentence is a canonical id in this graph - LEDGF/PSIP1, PAF1, KMT2A/MLL1, SEC - and the result arrives from the LEDGF side rather than from the KMT2A side.

Why that matters for this entry specifically. Route 1 here is that CK2 phosphorylates KMT2A at S136/S142 and raises LEDGF/p75 affinity, with the phospho-null fusion transformation-impaired. That establishes the phosphorylation and an affinity change. It does not establish what fails downstream when CK2 is inhibited, and the HIV work supplies exactly that in a human system: PAF1 stays, MLL1 does not arrive, and SEC does not arrive. It also converges with route 2 from a different direction - route 2's reported consequence of CK2 inhibition is that MLL chimeras are displaced from chromatin. Two independent routes, one shared endpoint: no CK2, no MLL on chromatin. The provirus is not a KMT2A fusion target and the inference is not automatic, but the convergence is worth recording and is currently recorded nowhere.

The limit that applies to all of it, and it is this gene's standing problem. data/symbol-collisions.json records that CSNK2A1 is an alias symbol of CSNK2A2, and that PubMed strips the apostrophe from CK2alpha' so that CK2alpha'[tiab] and CK2alpha[tiab] return the identical 792 records. Every result in this section was obtained with pan-catalytic-subunit chemistry (CX-4945, TBB, DMAT) or with subunit-nonspecific depletion, so none of it is attributable to CSNK2A1 rather than CSNK2A2. All queries here were run on HGNC symbols only. This is naming, not measurement, and the entry already says so for the other three routes.

Facts and figures
uniprotP68400
length AA391
gene symbolCSNK2A1
uniprot verifiedP68400 = CSK21_HUMAN, primary gene name CSNK2A1, 391 aa, casein kinase II subunit alpha, Homo sapiens (UniProt REST, read back 2026-09-19; primary gene name asserted, not a synonym match)
subnuclear localizationnucleolus, in a bimodal and prognostically meaningful way - present in 50.8% of lung adenocarcinomas and an independent predictor of worse recurrence-free survival (PMID 39513582). Not measured in any leukaemia. No speckle, nuclear-body or heterochromatin assignment for CSNK2A1 was located.
shown for fusionNO. CK2 appears three times in this graph's KMT2A biology - tightening the fusion's interface, phosphorylating LEDGF, and phosphorylating BRG1 during mitosis - and its localisation has never been measured in a KMT2A-rearranged cell. The bimodality is what makes this worth chasing: in lung adenocarcinoma roughly half of tumours have nucleolar CK2alpha and half do not, so if the same is true of leukaemic blasts, 'CK2 is present' is not a sufficient statement of where the kinase is.
cell context evidence classmaintains_a_state (both the compartment phenotype and the transcriptional state). NOT required_in - the knockout animals are viable with a mild baseline phenotype - and no cell-type-restricted expression is claimed.
cell context shown for fusionNO. Measured 2026-09-21: (CSNK2A1[tiab] OR "CK2alpha"[tiab] OR "casein kinase 2"[tiab] OR CK2[tiab]) AND (KMT2A[tiab] OR MLL[tiab]) returns 4 records, none of them a cell-type or cell-state experiment. Both cell-context results are normal mouse haematopoiesis; neither touches a KMT2A-rearranged cell. Zero relevant in titles and abstracts; supporting figures not searched.
cell context paralogue resolutionThe alias hazard is resolved here rather than waved at. This graph's registry records that CSNK2A1 is an ALIAS SYMBOL of CSNK2A2, so a source saying 'CK2alpha' has to be resolved before it can be attributed. The preprint names the deleted gene explicitly as Csnk2a1 and states that CK2alpha is the predominant catalytic subunit in haematopoietic cells; the Cells 2025 paper describes an alpha-specific conditional knockout distinguished from alpha-prime. Both therefore resolve to CSNK2A1 and not CSNK2A2. This does NOT extend to the chemistry: CX-4945, TBB and DMAT are pan-catalytic-subunit, so any CK2 inhibitor result in this entry remains unattributable between the two subunits.
cell context depmapNo selectivity. CSNK2A1 is -0.122 in the 11 KMT2A-rearranged lines against -0.100 in 43 other leukaemia lines, difference -0.022, rank 6,313 of 17,728 - effectively the middle of the distribution, on a pan-line mean of -0.115. Consistent with the mouse: CK2alpha is not a survival gene in blood cells.
epigenetic readoutTHE ONE GENUINE HISTONE WRITER IN THIS TRANCHE - AND IT IS NOT WHAT THE KINASE IS USUALLY WRITTEN UP FOR. Of fourteen entries amended in this tranche, thirteen write no histone mark. CK2 does: casein kinase 2 PHOSPHORYLATES HISTONE H4, and H4 phospho-serine-1 (S1P) strongly decreases the ability of the NuA4 H4/H2A acetyltransferase complex to acetylate H4 peptides, an inhibition that arginine-3 dimethylation alleviates; DNA-damaging agents raise H4 S1P with a concomitant fall in H4 acetylation, CK2 associates with the Rpd3 deacetylase complex, and ChIP links local H4 phosphorylation to H4 deacetylation during both transcription and DNA repair (PMID 16135807 - YEAST; NuA4 and Rpd3 are S. cerevisiae complexes; cited in zero entries of this corpus before 2026-09-21). So CK2 writes a phospho-mark that GATES AN ACETYL MARK - a phospho/acetyl switch on the H4 tail. SECOND, NON-HISTONE BUT MARK-LEVEL: CK2 phosphorylates the histone chaperone SPT6, and that phosphorylation is required to suppress histone H3 turnover and to inhibit cryptic sense and antisense transcription; CK2 depletion causes a replication-independent increase in H3K56 ACETYLATION and global activation of H3 turnover in coding regions (PMID 29905868 - YEAST, also uncited here). THIRD, AND THIS IS THE ON-BEAT ONE: CK2 LICENSES MLL1 AND SEC RECRUITMENT TO CHROMATIN THROUGH LEDGF. At the HIV provirus, LEDGF/p75 suppresses transcription during latency by recruiting the PAF1 complex and pausing Pol II; on latency reversal the MLL1 complex competitively DISPLACES PAF1 through a CASEIN-KINASE-II-DEPENDENT ASSOCIATION WITH LEDGF/p75, and depleting or pharmacologically inhibiting CKII prevents PAF1 dissociation and ABROGATES THE RECRUITMENT OF BOTH MLL1 AND THE SUPER ELONGATION COMPLEX (PMID 32426500, human; uncited here). That is the functional consequence this entry's route 1 - CK2 phosphorylating KMT2A S136/S142 to raise LEDGF affinity - did not have, arrived at independently from the LEDGF side. PARALOGUE AND SYMBOL LIMIT, per data/symbol-collisions.json: every result above is CK2-HOLOENZYME, obtained with pan-catalytic-subunit chemistry (CX-4945, TBB, DMAT) or subunit-nonspecific depletion, so NONE of it is attributable to CSNK2A1 rather than CSNK2A2 - CSNK2A1 is itself an alias symbol OF CSNK2A2, and PubMed strips the apostrophe from CK2alpha' so that CK2alpha and CK2alpha-prime return the same 792 records. Queried by HGNC symbol only. CAUSE OR CONSEQUENCE: for H4 S1P and for the MLL1/SEC recruitment the perturbations are genetic and pharmacological and the direction is established in those systems; neither has been measured in a KMT2A-rearranged cell.
paralogues
[
  "CSNK2A2 (CK2 alpha')",
  "CSNK2B (CK2 beta regulatory subunit)"
]
substrates in this graph
[
  {
    "substrate": "gene-kmt2a",
    "sites": "S136, S142",
    "fate": "retained"
  },
  {
    "substrate": "gene-kmt2a",
    "sites": "T2724, S2726",
    "fate": "lost"
  },
  {
    "substrate": "DEK",
    "sites": "S301/303/306/307",
    "fate": "not applicable"
  }
]
inhibitors
[
  "CX-4945 (silmitasertib)",
  "TTP22",
  "DMAT",
  "TBB"
]
localization
{
  "subcellular": "both nucleus and cytoplasm; CK2 is not compartment-restricted. The specific, measurable localisation claim is subnuclear rather than subcellular - see below.",
  "subnuclear": "NUCLEOLUS, as a discrete and clinically scored pool. Immunohistochemistry on 118 resected early-stage lung adenocarcinomas scored CK2alpha localisation and found nucleolar staining in 50.8% of tumours, independent of stage, histological type and grade, with nucleolar-positive patients having significantly worse recurrence-free survival and nucleolar CK2alpha an independent poor prognostic factor on multivariate analysis. The same group had previously linked CK2alpha localisation to prognosis in invasive breast cancer. The nucleolus is therefore a real, scoreable CK2alpha compartment - in solid tumours.",
  "how_shown": "immunohistochemistry scored for nucleolar versus non-nucleolar CK2alpha on 118 surgical specimens with survival follow-up (PMID 39513582)",
  "systems": "resected human lung adenocarcinoma; a previous report in invasive breast cancer. No haematopoietic or leukaemic system.",
  "node": "localization-nucleolus-kmt2a-fusion"
}
cell type
{
  "maintains_state_in": "The haematopoietic stem and early progenitor compartment. Vav-iCre conditional deletion of CK2alpha (Csnk2a1) in the mouse haematopoietic compartment gives viable animals with no dysmorphic features and a mild baseline phenotype, but a significant fall in total red blood cells and platelets, an enlarged spleen with extramedullary haematopoiesis, and EXPANSION of the HSC and early progenitor compartments - read by the authors as impaired proliferation and differentiation of those cells.",
  "not_required_for_survival": "Explicitly not. The animals live and the baseline phenotype is mild, so CK2alpha is a differentiation-and-output gene in haematopoiesis rather than a survival gene. Global deletion of the alpha or beta subunit is embryonically lethal, which is a different exposure and must not be quoted as a haematopoietic requirement.",
  "how_shown": "Vav-iCre conditional knockout mouse, blood counts, spleen histology, flow cytometry of HSC and progenitor compartments; single-cell RNA-seq of bone marrow and spleen in the companion preprint",
  "systems": "mouse; no human, no leukaemia, no KMT2A-rearranged material",
  "nodes": [
    "cell-type-hsc-bone-marrow",
    "cell-type-megakaryocyte"
  ]
}
cell state
{
  "maintains_state": "An immune-quiescent transcriptional state in the HSC. Single-cell RNA-seq of Csnk2a1 conditional-knockout mouse haematopoiesis found HSC ABUNDANCE unchanged but the HSC transcriptome the most strongly altered of any cell type, remodelled toward inflammatory and immune programmes, interferon signalling and antigen presentation, with Nfkb1, Rfx5, Hes1 and AP-1 family members as the regulatory hubs; S100a8 and S100a9 moved in OPPOSITE directions in bone marrow versus spleen HSCs, so the consequence is shaped by tissue context.",
  "reading": "CK2alpha loss does not remove the HSC; it moves it to an immune-primed state. That is a textbook maintains-a-state result, and it is the second time in this tranche that a gene with no viability phenotype turns out to hold a cell identity.",
  "nodes": [
    "cell-state-interferon-herv-mhc2"
  ]
}
depmap kmt2a selectivity
{
  "chronos_mean_kmt2a_rearranged": -0.1218,
  "chronos_mean_other_leukaemia": -0.0999,
  "difference": -0.0219,
  "pan_essentiality_all_lines": -0.1151,
  "selectivity_rank": 6313,
  "rank_denominator": 17728,
  "provenance": "Joined 2026-09-21 from analyses/depmap-kmt2a-selective-scan-full (generated 2026-09-19T10:12:06Z), the project's own genome-wide scan over 11 KMT2A-rearranged and 43 other leukaemia DepMap lines. NOT re-fetched and NOT independent of that scan: every caveat it carries applies here, including that the KMT2A-rearranged arm is a small number of lines drawn from nine individuals, that the rank denominator is 17,728 genes, and that a difference is not a dependency. Added because only 26 of 308 gene entries carried this number, which made any KMT2A-selectivity question answerable for 8% of the corpus.",
  "how_to_read": "A negative difference means MORE required in KMT2A-rearranged lines than in other leukaemia lines. A rank near 17,728 is ANTI-selective. Neither is evidence of a fusion-driven dependency on its own: check pan_essentiality_all_lines before calling anything selective, and note that a paralogue-redundant gene can score near zero for reasons that have nothing to do with this disease - see ana-the-paralogue-redundancy-excuse-is-not-tested-and-my-proxy-decided-the-answer."
}
How this could bite us back
  • CK2 is essential in most cell types and its inhibition has broad effects; a window argued from any single substrate is not a window, and CX-4945 has documented CK2-independent activity.
  • Route 2 requires a functional wild-type KMT2A allele to work. That makes it a therapy whose mechanism can be lost by a single additional genomic event, and one that would be invisible in a cell-line panel that happens to retain the allele.
  • Three CK2 routes converging on the same clinical recommendation is a resistance liability as much as a validation: a cell that upregulates CK2 or bypasses one substrate keeps the others, and dose escalation will hit the pleiotropic kinase rather than the leukaemic mechanism.
  • CK2 supports normal haematopoietic and neuronal function; paediatric exposure to a constitutive-kinase inhibitor is a developmental-toxicity question this graph has not asked.
  • Every CK2 inhibitor this graph names - CX-4945, TBB, DMAT - hits both catalytic subunits, and the two subunits differ specifically in subcellular localisation (PMID 38132153). So a CK2 inhibitor cannot be aimed at a compartment, and if the nucleolar pool (PMID 39513582) is the one driving ribosome biogenesis, on-target nucleolar stress is an unavoidable part of the dose rather than an off-target effect that could be engineered away.
  • ESSENTIAL-FUNCTION TOXICITY — Haematopoietic CK2alpha deletion in mice lowers red cells and platelets and produces extramedullary haematopoiesis. A CK2 inhibitor given to an infant whose marrow is already replaced by leukaemia and already being ablated by chemotherapy is being given into that phenotype. The mouse is reassuring on survival - the animals are viable with a mild baseline phenotype - and unreassuring on counts.
  • CX-4945 is presented in this entry as selectively depleting leukaemia stem cells while sparing normal HSPCs, and the epigenetic layer suggests where the unmeasured toxicity would be. CK2 restrains histone acetylation and histone H3 turnover and suppresses cryptic sense and antisense transcription; losing that restraint genome-wide is a chromatin-destabilising effect rather than a targeted one, and the readouts used to declare HSPC sparing are viability and engraftment, neither of which would see a heritable change in chromatin state. A washout-and-rechallenge design, like the one this graph already proposes for DOT1L, is the relevant control.
Still open
  1. In a single t(4;11) line, does CX-4945 loosen the fusion's LEDGF contact (route 1) before or after it stabilises wild-type MLL1 (route 2)? A dose- and time-resolved experiment would separate the two, and nobody has run it.
  2. Does CK2 inhibition combine with or antagonise menin inhibition? Route 1 predicts additivity on the clamp; route 2 predicts competition for the same chromatin sites, which could look like antagonism.
  3. Is CK2 activity itself elevated in KMT2A-rearranged leukaemia, or is the dependency about substrate availability? Neither study addresses this.
  4. Does the wild-type-allele mechanism survive in patients whose second KMT2A allele is deleted or silenced?
  5. CK2alpha loss remodels the normal HSC transcriptome toward interferon and antigen presentation, and this graph records interferon/MHC-II induction as a menin-inhibitor effect in KMT2A-rearranged leukaemia. Does CK2 inhibition raise MHC class II or interferon output in a KMT2A-rearranged blast, and if so does it add to or occlude the menin-inhibitor effect? Both reagents are available and the readout is a flow panel.
  6. The compartment study reports HSC expansion and the single-cell study reports unchanged HSC abundance in nominally the same model. Which is right, and does it matter for the therapeutic argument? A shared gating strategy across the two cohorts would settle it.
  7. Does CK2 inhibition abrogate KMT2A-FUSION and SEC recruitment to fusion target loci, as it abrogates MLL1 and SEC recruitment to the HIV provirus? Route 1 in this entry establishes that CK2 phosphorylation of KMT2A S136/S142 raises LEDGF affinity and that the phospho-null fusion is transformation-impaired; the HIV work supplies the missing downstream readout in a human system. A CX-4945 arm with fusion, menin, PAF1 and SEC ChIP at HOXA9 and MEIS1 in a KMT2A::AFF1 line would close the loop, and it would also test whether PAF1 displacement is part of the fusion's mechanism.
  8. Is H4 serine-1 phosphorylation present at KMT2A-fusion target loci, and does it restrain acetylation there? CK2 phosphorylates H4 and H4 S1P inhibits NuA4 acetylation of H4 in yeast, CK2 has a clinical-stage inhibitor, and this graph carries H4K16ac as a mark node. The measurement has never been made in any mammalian leukaemia, and H4S1ph is nearly unsearchable - CK2[tiab] AND H4S1ph[tiab] returns 0 - so it would not be found by a symbol or mark query.
  9. Can any of the four CK2 routes be assigned to CSNK2A1 rather than CSNK2A2? None can, on current tools: CX-4945, TBB and DMAT are pan-catalytic-subunit, CSNK2A1 is an alias symbol of CSNK2A2, and PubMed merges CK2alpha with CK2alpha-prime by stripping the apostrophe. An isoform-selective degrader, or a paralogue-resolved CRISPR pair in a KMT2A-rearranged line, is the only way this becomes a measurement rather than a naming convention.
Every source behind this entry (10)
  1. E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    UniProtKB annotates KMT2A S136 and S142 as phosphoserine 'by CK2' with experimental evidence traced to the LEDGF/p75 IBD interactome study, in which IBM phosphorylation switches partner affinity and MLL1 S136A/S142A impairs fusion-driven transformation.

    PMID 29997176DOI
    What this evidence cannot show

    Read as abstract plus PMC HTML full text on 2026-09-19. The explicit CK2-phosphorylates-MLL1 assignment is UniProt's curation of this paper; the paper's own directly reported CK2 treatment experiment is on POGZ, with the MLL1 evidence being the site mutants and the transformation assay. That is a real gap between what the paper shows and what the database asserts, and it is stated here rather than smoothed over.

    Evidence that argues against this (1)
    • CK2 is a constitutively active kinase with several hundred reported substrates; assigning a phenotype to one of them requires site-resolved rescue, and the transformation experiment used phospho-null and phosphomimetic KMT2A mutants rather than CK2 perturbation, so it demonstrates that the sites matter without demonstrating that CK2 is the kinase that writes them in that setting. UniProt Consortium 2026
  2. E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    Casein kinase II phosphorylates MLL1 proximal to its taspase1 cleavage site and facilitates cleavage; CKII knockdown or pharmacological inhibition blocks MLL1 processing, increases full-length MLL1, increases MLL1 chromatin occupancy genome-wide, displaces MLL-AFF1, AFF4 and ELL2 from chromatin in SEM cells, and CX-4945 delays leukaemic progression and improves survival in an MLL-AF9 mouse model.

    PMID 30573454DOI
    What this evidence cannot show

    Full text read via Europe PMC fullTextXML (PMC6317322). Much of the mechanistic work is in 293T cells; the leukaemia work is in SEM/MV4-11 and a mouse MLL-AF9 model. CX-4945 has substantial CK2-independent activity reported elsewhere, and the survival benefit is a delay rather than a cure. Residue numbers T2724/S2726 are our mapping of the paper's peptide onto Q03164.

    Evidence that argues against this (1)
    • This mechanism works through the wild-type allele, not through the fusion: the paper explicitly reports that CX-4945 increased full-length MLL1 without altering MLL-AF4 protein levels. Any benefit therefore depends on a functional second KMT2A allele, which is the opposite of a fusion-directed therapy and would be expected to fail where the wild-type allele is lost or silenced. Zhao Z et al. 2019
  3. E3moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

    CK2 directly phosphorylates DEK at Ser301/303/306/307, a modification required for the DEK-GABPA interaction that drives PBX3 expression; blocking it by mutation or with CX-4945 selectively depletes leukaemia stem cells while sparing normal HSPCs in MLL-AF9 and HOXA9/MEIS1 models and patient-derived xenografts, and CX-4945 synergises with venetoclax.

    PMID 42378254DOI
    What this evidence cannot show

    Abstract read 2026-09-19; full text not retrieved. A 2026 publication with no independent replication yet. DEK is not a KMT2A contact, so this is a third CK2 route into the same disease rather than a third route into the same protein, and the LSC-sparing claim rests on the authors' own comparison of leukaemic and normal progenitors.

    Evidence that argues against this (1)
    • A phosphoproteomic difference between leukaemia stem cells and normal HSPCs identifies where signalling differs, not where a therapeutic window is; the graph's own standing caution is that a selectivity claim drawn from an index-versus-comparator comparison must be checked for lineage and floor effects before it is treated as a window, and no such check is available from the abstract. Casado P et al. 2023
  4. E-ck2-alias-trapestablished confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database

    HGNC:2459 (CSNK2A2, UniProt P19784, 350 aa) carries CSNK2A1 as an alias symbol, while CSNK2A1 is HGNC:2457 (UniProt P68400, 391 aa), so an alias-matching lookup for CSNK2A1 can return CSNK2A2. The CK2 inhibitors CX-4945, TBB and DMAT are pan-catalytic-subunit, so results obtained with them do not distinguish the two paralogues.

    What this evidence cannot show

    HGNC records fetched 2026-09-20 and the alias lists read back. The claim about inhibitor selectivity is a statement about the compounds' published profiles and is the reason the attribution fails; **it is not a claim that CSNK2A2 carries the biology**, which is equally unmeasured.

    Evidence that argues against this (1)
    • Pan-subunit tools are not useless: if both paralogues must be inhibited for the phenotype, a pan-subunit compound is the correct probe and a single-gene knockout would have been the misleading one. The defect is in the attribution to one symbol, not in the experiments. Broad Institute DepMap via Open Targets 2026
  5. E5moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

    CK2alpha has a discrete nucleolar pool that varies between tumours of the same type and carries prognostic information: nucleolar CK2alpha staining was present in 50.8% of 118 resected early-stage lung adenocarcinomas, was independent of pathological stage, histological type and grade, and was an independent predictor of worse recurrence-free survival on multivariate analysis alongside lymph node metastasis.

    Nucleolar casein kinase 2 alpha as a prognostic factor in patients with surgically resected early‑stage lung adenocarcinoma. · Muto S, Homma MK, Kiko Y, Ozaki Y, Watanabe M, Okabe N, Hamada K, Hashimoto Y, Suzuki H · Oncol Rep 2025
    PMID 39513582DOI
    What this evidence cannot show

    Single-centre retrospective immunohistochemistry in a solid tumour - not leukaemia, not haematopoietic, and per brief rule 73 no claim derived from it may open without saying so. IHC scoring of a subnuclear compartment is observer-dependent and the antibody clone is not recorded here. And the deeper problem is the identifier one below: an anti-'CK2alpha' antibody's paralogue specificity determines what this result is about.

    Evidence that argues against this (1)
    • Brief rule 63 and this graph's own collision register apply directly: CSNK2A1 is an alias symbol OF CSNK2A2, the two catalytic subunits share 90% identity across the catalytic domain, and their regulation differs specifically in SUBCELLULAR LOCALISATION, post-translational modification and aggregation. CK2alpha' is not a backup of CK2alpha on this axis. A nucleolar 'CK2alpha' stain is therefore a claim about whichever isoform the antibody sees, and every CK2 inhibitor in this graph - CX-4945, TBB, DMAT - is pan-catalytic-subunit, so no pharmacological experiment can attribute a compartment to CSNK2A1 rather than CSNK2A2. Montenarh M et al. 2023
  6. E6moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

    Conditional deletion of CK2alpha in the mouse haematopoietic compartment (Vav-iCre CK2alpha f/f) produces viable animals without dysmorphic features and with a mild baseline phenotype, but with a significant decrease in total red blood cells and platelets, an enlarged spleen showing extramedullary haematopoiesis, and expansion of the haematopoietic stem cell and early progenitor compartments, consistent with impaired proliferation and differentiation of those cells.

    CK2α Deletion in the Hematopoietic Compartment Shows a Mild Alteration in Terminally Differentiated Cells and the Expansion of Stem Cells. · Rajaiah R, Daniyal M, Shanmugam MP, Valensi H, Duke K, Mercer K, Klink M, Lanza M, Uzun Y, Huang S, Dovat S, Behura CG · Cells 2025
    PMID 40643484DOI
    What this evidence cannot show

    Mouse conditional knockout, reported by the authors as the generation and validation of a new model with a mild baseline phenotype; the functional interpretation of the HSC expansion is explicitly flagged by them as warranting further study. No human, no leukaemia and no KMT2A-rearranged arm. Constitutive lifelong deletion in the whole haematopoietic compartment is not the exposure a CK2 inhibitor would give. Read: abstract only.

    Evidence that argues against this (1)
    • The compartment phenotype and the transcriptional phenotype do not agree on whether HSC number changes: this study reports expansion of the HSC and early progenitor compartments, while the single-cell companion reports HSC ABUNDANCE comparable between control and knockout, with the change being transcriptional rather than numerical. Different cohorts, different readouts and different gating, and the disagreement is unresolved. Valensi H et al. 2026
  7. E7moderate confidence 3 of 5: One strong study, or several smaller consistent ones.preprint

    Single-cell RNA sequencing of bone marrow and spleen from mice with conditional Csnk2a1 knockout in the haematopoietic compartment found HSC abundance comparable to control but the HSC transcriptome the most strongly affected of any cell type, remodelled toward inflammatory and immune-associated programmes, interferon signalling and antigen presentation, with Nfkb1, Rfx5, Hes1 and AP-1 family members identified as regulatory hubs and with S100a8/S100a9 changing in opposite directions in bone marrow versus spleen HSCs.

    Lineage-specific CK2α deletion reshapes the transcriptome of hematopoietic stem cells toward an immune-primed state. · Valensi H, Rajaiah R, Shanmugam M, Muhammad D, Golla U, Mercer K, Karampuri A, Dovat S, Behura C, Uzun Y · bioRxiv 2026
    PMID 42039464DOI
    What this evidence cannot show

    A bioRxiv PREPRINT, not peer reviewed, and the graph should not upgrade this until it is. Mouse steady-state haematopoiesis with no leukaemia and no KMT2A-rearranged arm. The regulatory hubs come from a network inference rather than from perturbation, and the cell-cell communication analysis is a ligand-receptor inference from transcriptome data. Read: abstract only.

    Evidence that argues against this (1)
    • This graph's interferon cell-state entry records the interferon state in KMT2A-rearranged leukaemia as something a MENIN INHIBITOR creates through endogenous retrovirus induction and MHC class II, not as a baseline property. The CK2alpha result is an interferon-and-antigen-presentation programme in NORMAL mouse HSCs with no drug and no fusion, so the two states share vocabulary and not provenance - and the shared vocabulary is exactly the kind of thing that gets pooled by mistake. Rajaiah R et al. 2025
  8. E8moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

    CK2 IS A HISTONE KINASE AND ITS MARK GATES AN ACETYL MARK: casein kinase 2 phosphorylates histone H4, and phosphorylation of H4 serine 1 (S1P) strongly decreases the ability of the NuA4 H4/H2A acetyltransferase complex to acetylate H4 peptides, while arginine-3 dimethylation combined with S1P alleviates that repression; chromatin from cells treated with DNA-damage-inducing agents shows increased S1 phosphorylation with a concomitant decrease in H4 acetylation, CK2 associates with the Rpd3 deacetylase complex, and chromatin immunoprecipitation links local H4 phosphorylation to H4 deacetylation during both transcription and DNA repair, with time-course data placing H4 phosphorylation after NuA4 action during double-strand-break repair at the step of chromatin restoration.

    PMID 16135807DOI
    What this evidence cannot show

    Abstract read from the eutils record; full text not read in this tranche. YEAST: NuA4 and Rpd3 are Saccharomyces cerevisiae complexes, and the human counterparts (TIP60/KAT5 and the HDAC1/2-containing complexes) are not identical, so this is not a human result and rule 73 applies to any downstream claim. The HAT assays are on H4 TAIL PEPTIDES, not nucleosomes, and peptide HAT assays overstate the effect of tail modifications. The phosphorylation is attributed to 'casein kinase 2' as a holoenzyme; nothing here distinguishes CSNK2A1 from CSNK2A2 or from the yeast Cka1/Cka2 subunits. H4S1ph has essentially no searchable literature under that name - CK2[tiab] AND H4S1ph[tiab] returns 0 - so the mark is reachable only through phrase queries, which bounds any absence claim about it. Not cited anywhere in this corpus before 2026-09-21, verified with scripts/check-citation-reuse.py.

    Evidence that argues against this (1)
    • The writer activity should not displace what CK2 is actually established to do in this disease, which is phosphorylate non-histone proteins in the KMT2A machinery. Casein kinase II phosphorylates MLL1 proximal to its taspase1 cleavage site and facilitates cleavage; CKII knockdown or pharmacological inhibition blocks cleavage, stabilises full-length MLL1 and displaces MLL chimeras from chromatin, delaying leukaemia in an MLL-AF9 mouse - a human/mouse leukaemia result with a direct phenotype. Set against a yeast peptide-HAT experiment, the histone-writer activity is the weaker and more distant claim, and an entry that led with 'CK2 is a histone kinase' would misrepresent where the evidence in this disease actually sits. Zhao Z et al. 2019
  9. E9moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

    A NON-HISTONE CK2 SUBSTRATE WITH A MARK-LEVEL CONSEQUENCE: CK2 depletion from yeast cells causes a replication-independent increase in histone H3K56 acetylation and global activation of H3 turnover in coding regions, and CK2 inhibits global cryptic promoters driving both sense and antisense transcription; CK2 interacts with the major histone chaperone Spt6 and phosphorylates it in vivo and in vitro, and that phosphorylation is required for Spt6's cellular levels, for suppression of histone H3 turnover and for inhibition of spurious transcription.

    PMID 29905868DOI
    What this evidence cannot show

    Abstract read from the eutils record; full text not read. YEAST throughout, and Spt6 phosphosite-level detail is not in the abstract. The H3K56ac increase and H3 turnover activation follow CK2 DEPLETION, which is a broad perturbation of an essential kinase with hundreds of substrates, so attributing the chromatin phenotype to the Spt6 axis rests on the phosphosite mutants rather than on the depletion. Subunit-nonspecific: 'CK2' here is the holoenzyme and the yeast catalytic subunits are Cka1/Cka2, so nothing maps onto CSNK2A1 versus CSNK2A2. The human orthologue SUPT6H has an entry in this graph and no equivalent experiment. Not cited anywhere in this corpus before 2026-09-21.

    Evidence that argues against this (1)
    • The two yeast CK2 chromatin results point in opposite directions on acetylation and the entry should not average them. Here CK2 DEPLETION INCREASES H3K56 acetylation, implying CK2 normally restrains an acetyl mark; in the NuA4 work CK2's own H4 S1 phosphorylation DECREASES H4 acetylation and CK2 associates with a deacetylase complex, implying the same restraining direction but through a different substrate and a different histone. They are consistent in sign and incompatible in mechanism - one runs through a chaperone and histone turnover, the other through direct tail phosphorylation and HAT inhibition - and no experiment separates their contributions. Recorded as two routes, not one. Utley RT et al. 2005
  10. E10strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    CK2 LICENSES MLL1 AND SEC RECRUITMENT TO CHROMATIN THROUGH LEDGF, AND THE ENDPOINT CONVERGES WITH THIS ENTRY'S ROUTE 2: at the HIV provirus, LEDGF/p75 suppresses proviral transcription during latency via promoter-proximal pausing of Pol II by recruiting the PAF1 complex; following latency reversal the MLL1 complex competitively displaces PAF1 from the provirus through a casein-kinase-II-dependent association with LEDGF/p75; and depleting or pharmacologically inhibiting CKII prevents PAF1 dissociation and abrogates the recruitment of both MLL1 and the Super Elongation Complex, thereby impairing transcriptional reactivation.

    PMID 32426500DOI
    What this evidence cannot show

    Abstract read from the eutils record; full text not read in this tranche. THE SYSTEM IS AN INTEGRATED HIV PROVIRUS, not a KMT2A fusion target locus and not a leukaemia - LEDGF/p75 is the lentiviral integration tether, which is why it is studied there, and rule 73 applies: nothing here is index-fusion evidence. The MLL1 involved is WILD-TYPE MLL1, not a fusion protein, so the result bears on route 1's mechanism by analogy and not by demonstration. 'CKII-dependent' rests on depletion plus pharmacological inhibition, both subunit-nonspecific, so CSNK2A1 versus CSNK2A2 is not resolved. The abstract does not state which CKII inhibitor was used or at what selectivity. Not cited anywhere in this corpus before 2026-09-21.

    Evidence that argues against this (1)
    • The direction of the CK2-LEDGF dependency is established from the KMT2A side in this graph as an AFFINITY effect and not as a recruitment requirement, and the two are not the same claim. UniProt annotates KMT2A S136 and S142 as phosphoserine 'by CK2' with experimental evidence traced to the LEDGF/p75 IBD interactome study, and the phospho-null fusion is transformation-impaired - which establishes that the phosphorylation matters for transformation without establishing that CK2 inhibition strips MLL1 off chromatin. Conversely the HIV work establishes the recruitment failure without touching a fusion. Neither experiment has been done in the other's system, so the convergence recorded in this section is an inference across two systems and is labelled as one. Sharma S et al. 2018
Connected to