DPY30 — the smallest COMPASS subunit, and the only one that behaves like a selective dependency

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.

• Essential-function toxicity: conditional Dpy30 loss in adult mouse blood causes severe pancytopenia with a differentiation block in HSCs. That is the same lesion the leukemia has, installed iatrogenically in normal blood, and it is worse than anything reported for menin loss.
• Compensation: DPY30 is shared by all six SET1/MLL complexes, so partial inhibition is buffered by whichever complex is least dependent on precise targeting, and the residual H3K4me2 after DPY30 loss sits exactly at highly transcribed genes, which is where an oncogenic program already is.
• Delivery: there is no pocket. A 99-residue protein whose function is one extended helical interface is a degrader problem, and a degrader of a subunit shared by six complexes has no selectivity handle.
• Safety-triage failure mode: DPY30 was called GREEN on LOEUF 0.73, a statistic that cannot resolve constraint in a gene this short. Any pipeline that promotes targets on LOEUF alone will keep making this mistake on small proteins.

1. Why does ASH2L fail the co-dependency test when it has the same essentiality distribution as DPY30? A direct test is to ask whether an ASH2L separation-of-function mutant that keeps RBBP5 scaffolding but loses the DPY30 interface phenocopies DPY30 loss in a KMT2A::AFF1 line.
2. Is the DPY30-MEN1 correlation a lineage artefact? Repeating the correlation within B-ALL lines and within AML lines separately would show whether it survives a narrower lineage window or evaporates.
3. Does DPY30 loss reduce H3K4me2/me3 preferentially at KMT2A-fusion target loci rather than globally? That is the direct test of the 'positioning not catalysis' explanation and it is a single CUT&RUN experiment in SEM cells.
4. DPY30's most depleted leukemia lines are NPM1-mutant AML rather than KMT2A-rearranged. Does the DPY30 requirement in fact track the menin-dependent state generally, of which KMT2A rearrangement is only one entry point?
5. What is the real human constraint on DPY30? LOEUF cannot answer it at this gene length. A targeted look at biallelic DPY30 variants in developmental-disorder cohorts would.

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

   DPY30 is not DepMap common-essential (mean Chronos -0.390 across 1,258 lines, -0.455 across 57 leukemia lines, 31.5% of lines depleted); gnomAD reports pLI 0.512 and LOEUF 0.731 with 6 observed against 16.2 expected loss-of-function variants; HPA reports low tissue specificity with bone marrow 29.6 nTPM.

   DepMap CRISPR (Chronos) gene effect via the Open Targets Platform GraphQL API; gnomAD v4 constraint via the gnomAD GraphQL API; Human Protein Atlas consensus tissue nTPM via the HPA API · Broad Institute DepMap; gnomAD; Human Protein Atlas; Open Targets Platform · Databases 2026
   ▸What this evidence cannot show

   Per-line Chronos values were retrieved through Open Targets and the means computed by us; the DepMap release version is not stated in the API response and remains unverified. The leukemia subset is 57 lines of any leukemia type and contains no infant KMT2A-rearranged line. LOEUF for a 99-residue protein is statistically underpowered and should not be read as evidence of tolerance to loss.

   ▸Evidence that argues against this (1)

   • A Chronos mean mixed across a KMT2A-rearranged subset and a comparator arm must be lineage-split before it is stated: the index arm is myeloid-enriched, so a difference between the two mixes fusion status with lineage. Ten or eleven index lines and four KMT2A::AFF1 lines also put wide intervals on every figure quoted. A pLI of 0.512 and LOEUF 0.731 also put DPY30 in the band where constraint metrics are least informative, since a small gene accumulates few expected loss-of-function variants and its interval is correspondingly wide. gnomAD, Broad Institute 2026
2. E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

   Conditional Dpy30 deletion in the adult mouse haematopoietic system causes severe pancytopenia with accumulation of HSCs and early progenitors that are defective in multilineage reconstitution, and Dpy30-deficient HSCs fail to sustain long term in mixed chimeras.

   Dpy30 is critical for maintaining the identity and function of adult hematopoietic stem cells. · Yang Z, Shah K, Khodadadi-Jamayran A, Jiang H · Journal of Experimental Medicine 2016
   PMID 27647347DOI
   ▸What this evidence cannot show

   Complete conditional gene deletion in mouse, which models total protein removal rather than partial reversible pharmacological inhibition, and mouse haematopoiesis differs from infant human haematopoiesis in ways that matter for this project.

   ▸Evidence that argues against this (1)

   • This is a mouse knockout describing what the gene does in normal development or normal haematopoiesis, over the animal's whole life, with the protein entirely absent. A drug gives partial reversible inhibition in an adult for weeks. The experiment bounds the biological requirement and not the therapeutic window. Severe pancytopenia with defective multilineage reconstitution is the phenotype a therapeutic window would have to avoid, and it appears at complete deletion in adult haematopoiesis — the closest model available to an adult on a drug. Nature 2012
3. E3moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

   DPY30 is required for the growth of several MLL1-fusion-driven leukemia cell lines, and DPY30 knockdown in human CD34+ progenitors impairs proliferation and myelomonocytic differentiation while promoting haemoglobin production.

   The DPY30 subunit in SET1/MLL complexes regulates the proliferation and differentiation of hematopoietic progenitor cells. · Yang Z, Augustin J, Chang C, Hu J, Shah K, Chang CW, Townes T, Jiang H · Blood 2014
   PMID 25139354DOI
   ▸What this evidence cannot show

   The MLL1-fusion leukemia line result is a short shRNA knockdown experiment reported at the end of a paper whose main subject is normal progenitors; the number of lines is small and no KMT2A::AFF1 infant model is included. Knockdown of a shared complex subunit does not distinguish a fusion-specific requirement from a general H3K4 requirement.

   ▸Evidence that argues against this (1)

   • Required for the growth of several MLL1-fusion lines AND impairing proliferation and differentiation of normal human CD34+ progenitors is a two-sided result that the entry reports and does not resolve. The normal-progenitor half is the bite-back. Oncofusion knowledge graph (internal analysis) 2026
4. E4moderate confidence 3 of 5: One strong study, or several smaller consistent ones.dataset

   Under an identical leukemia-restricted protocol across 54 DepMap leukemia lines, DPY30 correlates with MEN1 at r = 0.446 (95% CI 0.142 to 0.658, permutation p = 0.001, Holm p = 0.015) while the other three WRAD subunits do not: WDR5 r = 0.004, RBBP5 r = -0.081, ASH2L r = -0.153, all with confidence intervals crossing zero.

   Do the other WRAD subunits behave like DPY30? Leukemia-restricted co-dependency with bootstrap CI and permutation test; script and results in analyses/wrad-discrimination/ · Broad Institute DepMap; Open Targets Platform; analysis run by this project · Database and derived analysis 2026
   ▸What this evidence cannot show

   Our own computation, not peer reviewed. 54 leukemia lines is a small sample and the bootstrap interval for DPY30 is wide. Correlation across cell lines does not establish physical interaction or shared pathway and can reflect shared lineage restriction; no lineage adjustment was made within the leukemia subset. The DepMap release version is not reported by the API.

   ▸Evidence that argues against this (2)

   • A co-dependency correlation can be produced entirely by two genes being required in the same lineage rather than in the same process, and cell-line panels drift genetically and transcriptionally in culture. Ben-David U et al. 2018
   • This project's own deeper re-analysis withdraws the DPY30 result that this evidence item rests on. Against variance-matched nulls drawn from 2,743 usable random protein-coding genes across the same 54 leukemia lines, the Holm-corrected survivors are KMT2A (r 0.783), PSIP1 (0.686), DOT1L (0.681), FLT3 (0.645), MEF2C (0.526) and KAT7 (0.517). DPY30 is explicitly among those that do NOT clear correction, alongside MEIS1 (Holm 0.067), MLLT1 and BCL2. The earlier 16-gene panel gave DPY30 Holm p = 0.015 using an 'exceeds the maximum of the null' criterion that the deep-pool run shows to be unstable by construction, because the null maximum rises mechanically as the pool grows. The DPY30-versus-other-WRAD-subunits contrast is therefore a contrast between one uncorrected near-signal and three nulls, not between a survivor and three failures. Broad Institute DepMap; Open Targets Platform; analysis run by this project 2026
5. E5established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database

   DPY30 homodimerises and docks the ASH2L SDI motif at 2:1 stoichiometry, and DPY30 is resolved as part of the MLL1 core complex bound to the nucleosome.

   RCSB PDB entries 3G36 (human DPY-30-like C-terminal domain, 1.20 A), 4RIQ (DPY-30 dimerisation/docking domain with the Ash2L Sdc1-DPY-30 interaction motif, 2.23 A), 6E2H (Ash2L SPRY+SDI with full-length DPY-30, 2.24 A), 6PWV (cryo-EM MLL1 core complex on the nucleosome, 6.2 A) and 7UD5 (MLL1-WRAD with an H2B-ubiquitinated nucleosome, 4.25 A), each verified against the RCSB data API · RCSB Protein Data Bank · RCSB PDB 2026
   ▸What this evidence cannot show

   6PWV at 6.2 A resolution cannot place side chains, so interface residue assignments from that entry are unreliable. We verified that each accession exists, its title, method and resolution; we did not independently re-refine or re-interpret the models.

   ▸Evidence that argues against this (1)

   • Counting PDB entries measures how tractable a protein was to CRYSTALLISE, which correlates with being small, ordered and well-behaved rather than with being druggable. What decides design value is whether the deposited density covers the site that matters, at a resolution that supports modelling, in the relevant state — and this graph has repeatedly found structures described in the direction that overstates what is resolved. A 2:1 homodimer docking one motif is an architecture; being resolved within the MLL1 core complex on a nucleosome is a reconstitution. Neither reports stoichiometry in a leukaemia cell. RCSB PDB 2026
6. E6strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

   RBBP5 directly binds the H2B-conjugated ubiquitin and orients the MLL complex on the nucleosome, and WDR5 holds the catalytic subunit into the module through the conserved Win motif, making both structurally load-bearing for catalysis across the SET1 family.

   Structural basis of nucleosome recognition and modification by MLL methyltransferases. · Xue H, Yao T, Cao M, Zhu G, Li Y, Yuan G, Chen Y, Lei M, Huang J · Nature 2019
   PMID 31485071DOI
   ▸What this evidence cannot show

   Cryo-EM of reconstituted complexes on designed nucleosomes; the in-cell stoichiometry and the contribution of subunits that are poorly resolved in the maps, including DPY30, are not directly addressed. The inference that this explains DepMap pan-essentiality is ours, not the authors'.

   ▸Evidence that argues against this (1)

   • WDR5 is not required in the same way by every SET1-family complex: MLL3 interacts with RAD and shows enhanced activity in the absence of WDR5, so 'WDR5 is structurally required' is true of MLL1 and SETD1A but not uniformly across the family. Alicea-Velazquez NL et al. 2016
7. E7strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

   DPY30 contributes only modestly to positive heterotypic allosteric regulation of KMT2 methyltransferase activity in vitro, while the extensive DPY30-ASH2L interface is critical for correct genome-wide placement of H3K4me2 and H3K4me3, with residual H3K4me2 after DPY30 loss concentrated at highly transcribed genes.

   Structural Analysis of the Ash2L/Dpy-30 Complex Reveals a Heterogeneity in H3K4 Methylation. · Haddad JF, Yang Y, Takahashi YH, Joshi M, Chaudhary N, Woodfin AR, Benyoucef A, Yeung S, Brunzelle JS, Skiniotis G, Brand M, Shilatifard A, Couture JF · Structure 2018
   PMID 30270175DOI
   ▸What this evidence cannot show

   In vitro methyltransferase assays and genome-wide profiling in cell lines, not in leukemia models. The paper does not test KMT2A fusions, and the 'positioning not catalysis' framing we build on it is our inference. In vitro allostery assays are sensitive to complex assembly conditions, which are themselves temperature- and concentration-dependent for this complex.

   ▸Evidence that argues against this (2)

   • Reconstituted MLL1 core complex assembly is strongly temperature- and concentration-dependent and the disassembled state is favoured at physiological temperature, so in vitro conclusions about the relative contribution of individual WRAD subunits may not transfer to the cell. Namitz KEW et al. 2023
   • This is the decisive objection to the whole WRAD thread and it should be read before any of it. The SET domain of MLL1 -- the catalytic activity that WRAD exists to stimulate -- is dispensable for maintaining haematopoietic stem cells AND for supporting leukemogenesis driven by the MLL-AF9 fusion oncoprotein in mouse. On Mll1 deletion, what is lost at MLL1 target genes is H4K16 acetylation, and inhibiting SIRT1 is sufficient to restore both the acetylation and target-gene expression. The conclusion the authors draw is that recruited MOF activity, not MLL1's intrinsic methyltransferase activity, is what maintains these genes. If H3K4 methylation by MLL1 is not required for MLL-fusion leukemia, then the subunit whose job is to enable that methylation is not a target in this disease, whatever its biochemistry. Mishra BP et al. 2014
8. E8moderate confidence 3 of 5: One strong study, or several smaller consistent ones.dataset

   This project's own target-safety triage placed DPY30 GREEN on the declared rule (not common-essential, LOEUF above 0.35, mean gene effect above -0.5), with codependency r 0.446, mean gene effect -0.3898 and LOEUF 0.731.

   Target safety triage, analyses/target-safety-triage/ (this project, cycle 9) · Oncofusion Knowledge Graph project · Derived analysis 2026
   ▸What this evidence cannot show

   The triage rule uses LOEUF as a tolerance proxy without a power check. For DPY30 that proxy is uninformative because only 16.2 loss-of-function variants are expected in a 99-residue gene. The GREEN call should be downgraded to UNKNOWN on the constraint axis, and the mouse conditional-knockout phenotype is a toxicity signal the triage rule does not see at all.

   ▸Evidence that argues against this (1)

   • A GREEN call on this project's own declared triage rule is only as good as the rule, and the rule's thresholds — not common-essential, LOEUF above 0.35, mean gene effect above −0.5 — are cut-points on continuous variables chosen without calibration. The mouse pancytopenia in this same entry is the evidence the rule does not see. Statistics in Medicine 2006
9. E_yokoyama2004strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

   Biochemical purification of MLL identified a SET1-like complex containing ASH2L, WDR5, RBBP5, HCF-1, HCF-2 and menin. Oncogenic mutant (fusion) forms of MLL retain the ability to interact with menin but not the other identified complex components.

   Leukemia proto-oncoprotein MLL forms a SET1-like histone methyltransferase complex with menin to regulate Hox gene expression · Yokoyama A, Wang Z, Wysocka J, Sanyal M, Aufiero DJ, Kitabayashi I, Herr W, Cleary ML · Molecular and Cellular Biology 24:5639-49 2004
   PMID 15199122DOI
   ▸What this evidence cannot show

   Read as abstract only by the coordinator; the interactome beat reports reading the full text via PMC480881. 2004 affinity-purification mass spectrometry, at depths well below current practice, so a negative here is a failure to detect rather than a demonstration of absence. No proximity-labelling study of any KMT2A fusion has ever been published, so this negative has never been re-tested with a method capable of catching transient or substoichiometric contacts. **The founding negative is an inference from a deletion mutant, not a measured fusion co-immunoprecipitation.** Read in full 2026-09-20: WRAD subunits were mapped using a C-terminal deletion construct of otherwise wild-type MLL in 293 cells. The paper's only fusion experiments blot for HIS, MLL-C, HCF-1, menin, MLL-N and ENL — **no WRAD subunit is blotted on any fusion anywhere in it**. It is a measured negative for HCF-1 only. **And this paper is out of scope for DPY30 entirely**: the string 'DPY' appears zero times in its full text, and DPY30 was not assigned to the human SET1-like complexes until 2007. DPY30 also binds ASH2L rather than KMT2A directly, so the 2004 result could not cover it even in principle.

   ▸Evidence that argues against this (4)

   • A 2004 co-purification negative is weak evidence of non-interaction. Modern proximity labelling routinely recovers contacts that affinity purification misses, particularly transient ones, and the entire composition literature for this fusion rests on stable-complex purifications from 2004-2012. The claim here should be read as 'not detected as a stable complex member', not 'does not interact'. Yokoyama A et al. 2004
   • CORRECTED 2026-09-21 (cycle 44), and the previous version of this note was too strong. It said "zero published experiments test WRAD-versus-KMT2A-fusion association in either direction" and cited PMID 15199122 for that - a 2004 paper, which cannot establish an absence as of today, in a note whose own next sentence warned against citing that paper for more than it shows. Re-bounded by query today: the subunit-by-fusion co-occurrence search is NOT empty (WDR5 26 records, ASH2L 6, DPY30 3, RBBP5 2), and reading all 26 WDR5 titles returns two that genuinely bear on the question. PMID 27462455 shows MLL1 and MLL-AF9 occupy DISTINCT chromatin regions and attributes MLL1's enhancer recruitment to a C-terminal WDR5 interaction - hedged as "probably" by its own authors - which is an experiment discriminating wild-type from fusion with WDR5 as the proposed discriminator. PMID 36402263 shows genetic Wdr5 deletion attenuates MLL-AF9 leukemogenesis, acting through ribosomal genes rather than through fusion binding. THE SURVIVING ABSENCE IS NARROWER AND STILL WORTH STATING: no direct binding assay - co-IP, structure or proximity labelling - has tested whether WRAD associates with a KMT2A FUSION protein. The two experiments above are genomic occupancy and genetic requirement, neither of which is a binding measurement. The structural argument for exclusion is unchanged, because the WRAD contacts sit thousands of residues past the breakpoint. What this graph must not do is cite a 2004 deletion-mutant result as if it were a fusion co-IP negative. PubMed (bounded query, counts and date recorded) 2026
   • The nearest thing to a published test, and it cuts against the absence being total: MLL1 and MLL-AF9 occupy distinct chromatin regions, and the authors attribute MLL1's enhancer recruitment to a C-terminal WDR5 interaction that the fusion lacks. It is an occupancy experiment, not a binding assay, and the WDR5 attribution is hedged in the source. Xu J et al. 2016
   • And a functional requirement in the fusion context: genetic Wdr5 deletion attenuates MLL-AF9 leukemogenesis in vivo. That WDR5 is REQUIRED where the fusion drives disease does not show WRAD binds the fusion - the reported mechanism is ribosomal gene regulation - but it makes 'WRAD is irrelevant to the fusion' untenable. Liu L et al. 2023
10. E_thielstrong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    Menin recruits both wild-type MLL and the oncogenic MLL-AF9 fusion protein to HOX gene loci to activate transcription, and the wild-type Mll allele is required for MLL-AF9-induced leukemogenesis and for maintenance of MLL-AF9-transformed cells.

    MLL-AF9-induced leukemogenesis requires coexpression of the wild-type Mll allele · Thiel AT, Blessington P, Zou T, Feather D, Wu X, Yan J, Zhang H, Liu Z, Ernst P, Koretzky GA, Hua X · Cancer Cell 17:148-59 2010
    PMID 20159607DOI
    ▸What this evidence cannot show

    Read as abstract only. Mouse genetics with a retroviral MLL-AF9 model; the requirement is shown for MLL-AF9, not for KMT2A::AFF1, and this graph has repeatedly found partner-specific differences. 'Required' here is allele-level and does not identify which wild-type MLL module carries the requirement.

    ▸Evidence that argues against this (1)

    • A wild-type-allele requirement shown for MLL-AF9 in mouse need not hold for KMT2A::AFF1 in infant B-ALL, and this project has already been burned generalising across partners. The claim should be carried as strong for MLL-AF9 and untested for the fusion this project is about. Thiel AT et al. 2010
11. E_milnestrong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    The essential MLL1 recruitment domain includes both the CXXC domain and the PHD fingers and is controlled by interactions with the PAF elongation complex and H3K4me2/3. MLL1 fusion proteins lack the PHD fingers and require prebinding of a wild-type MLL1 complex, together with CXXC recognition of DNA, for stable HoxA9 association.

    Multiple interactions recruit MLL1 and MLL1 fusion proteins to the HOXA9 locus in leukemogenesis · Milne TA, Kim J, Wang GG, Stadler SC, Basrur V, Whitcomb SJ, Wang Z, Ruthenburg AJ, Elenitoba-Johnson KSJ, Roeder RG, Allis CD · Molecular Cell 38:853-63 2010
    PMID 20541448DOI
    ▸What this evidence cannot show

    Read as abstract only. Establishes a trans requirement at HoxA9 specifically; whether it generalises to the whole fusion target set is not addressed in the abstract. The authors themselves frame the distinct recruitment mechanisms as a therapeutic opportunity rather than an obstacle.

    ▸Evidence that argues against this (1)

    • If the fusion needs the wild-type complex prebound, then wild-type MLL1 recruitment is a potential target — but it is also a normal function in every cell that expresses MLL1, so the therapeutic window argument that applies to menin does not automatically transfer here. Milne TA et al. 2010
12. E_caostrong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    MM-401 inhibits MLL1 activity by blocking the MLL1-WDR5 interaction and preventing complex assembly, without affecting other MLL-family methyltransferases, and specifically blocks proliferation of MLL cells by inducing cell-cycle arrest, apoptosis and myeloid differentiation without general toxicity to normal bone marrow cells or non-MLL cells.

    Targeting MLL1 H3K4 methyltransferase activity in mixed-lineage leukemia · Cao F, Townsend EC, Karatas H, Xu J, Li L, Lee S, Liu L, Chen Y, et al. (Dou Y, senior author) · Molecular Cell 53:247-61 2014
    PMID 24389101DOI
    ▸What this evidence cannot show

    Read as abstract only. A peptidomimetic tool compound characterised in cells; no clinical or in vivo efficacy is claimed in the abstract, and 'MLL cells' spans MLL-rearranged lines generally rather than KMT2A::AFF1 specifically. SUPERSEDED IN PART (2026-09-19): MM-401's stated mechanism is inhibition of MLL1 methyltransferase activity, but WIN-site inhibition in MLL-rearranged cells is reported to act through ribosome protein gene suppression and p53 activation (PMIDs 30865883, 38682900). This citation should no longer be used as evidence that a wild-type-only MLL1 interface is druggable through MLL1 biology.

    ▸Evidence that argues against this (1)

    • MM-401's stated mechanism is inhibition of MLL1 H3K4 methyltransferase activity, yet the MLL1 SET domain is reported dispensable for maintaining HSCs and for MLL-AF9-driven leukemogenesis (Mishra 2014, PMID 24813891), which this graph also carries. Either the selectivity of MM-401 runs through something other than catalysis, or the two results are in tension. The graph should hold both rather than choose. Mishra BP et al. 2014
13. E_ahostrong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    Potent WIN-site inhibitors show that the WIN site links WDR5 to chromatin at a small cohort of loci including a specific subset of ribosome protein genes; the inhibitors rapidly displace WDR5 from chromatin and decrease expression of those genes, causing translational inhibition, nucleolar stress and p53 induction.

    Displacement of WDR5 from Chromatin by a WIN Site Inhibitor with Picomolar Affinity · Aho ER, Wang J, Gogliotti RD, Howard GC, Phan J, Acharya P, et al. (Tansey WP, senior author) · Cell Reports 26:2916-2928.e13 2019
    PMID 30865883DOI
    ▸What this evidence cannot show

    Read as abstract only. Establishes the mechanism of small-molecule WIN-site inhibitors across cancer contexts; it is not specific to KMT2A::AFF1 and does not directly test the peptidomimetic MM-401.

    ▸Evidence that argues against this (1)

    • MM-401 and the small-molecule WIN-site inhibitors are different chemotypes, and this paper does not test MM-401. It remains formally possible that the peptidomimetic acts differently. But the burden has shifted: the mechanism cannot now be assumed to be MLL1-programme disruption without direct evidence, and this graph previously assumed exactly that. Howard GC et al. 2024
14. E_howardstrong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    In MLL-rearranged cancer cells, WDR5 WIN-site inhibition acts by suppressing ribosome protein gene transcription, inducing nucleolar stress and activating p53, with depletion of the ribosome stock, a broad modest translational choke, and alternative-splicing changes that inactivate the p53 antagonist MDM4. The authors state that peptidomimetic WIN-site inhibitors were originally proposed to act via genes connected to haematopoietic stem cell expansion, and that their data show otherwise. WIN-site inhibitors are synergistic with venetoclax and BET-bromodomain inhibitors.

    Ribosome subunit attrition and activation of the p53-MDM4 axis dominate the response of MLL-rearranged cancer cells to WDR5 WIN site inhibition · Howard GC, Wang J, Rose KL, Jones C, Patel P, Tsui T, et al. (Tansey WP, senior author) · eLife 12:e90683 2024
    PMID 38682900DOI
    ▸What this evidence cannot show

    Read as abstract only. Cell-line work in MLL-rearranged lines; the p53 dependence is inferred from the mechanism rather than demonstrated across TP53-mutant and wild-type isogenic pairs in the abstract. Does not address KMT2A::AFF1 infant B-ALL specifically.

    ▸Evidence that argues against this (1)

    • A p53/ribosome mechanism does not make WIN-site inhibition useless in MLL-rearranged leukaemia - the same paper reports synergy with venetoclax and BET inhibitors, and a ribosome-directed strategy is novel in its own right. What it removes is the specific inference this graph drew: that MM-401 demonstrates an absent-from-the-fusion interface can be drugged through MLL1 biology. Howard GC et al. 2024

• ← part ofThe leukaemia term hid the answer for DPY30 — a rule-77 demonstration with clean rule-67 controls

  the bounded cell-state measurement for this gene

• ← depends onBone marrow haematopoietic stem cell — the postnatal comparator, and why the same fusion behaves differently here

  Dpy30 is critical for maintaining the identity and function of ADULT murine haematopoietic stem cells

• regulates →The stem-like self-renewal state — a limited stem-cell module grafted onto a committed cell

  loss collapses adult HSC identity and self-maintenance (murine, normal haematopoiesis)

• part of →KMT2A / MLL1 — the gene that breaks

  WRAD core subunit of the MLL1/KMT2A methyltransferase complex

• co-occurs with →MEN1 / menin — the adaptor the fusion cannot work without

  leukemia co-dependency r=0.45 (54 DepMap lines, survives Holm)

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

  same WRAD module; WDR5 is pan-essential and uncorrelated with MEN1

• ← supportsDo the other WRAD subunits behave like DPY30? (run, not planned)

  DPY30 is the only WRAD subunit surviving correction

• part of →Mechanism class 1 — chromatin and elongation hijack (KMT2A-r, NUP98-r)

  shared subunit of the H3K4 machinery the fusion works through

• ← bindsASH2L — the WRAD subunit that has DPY30's essentiality profile and none of its co-dependency signal

  SPRY/SDI surface docks the DPY30 homodimer at 2:1; crystal 4RIQ

• ← related toDPY30 aims the H3K4 machinery while ASH2L powers it, which is why only DPY30 tracks menin

  explains why DPY30 alone among WRAD tracks menin

• ← related toFirst systematic enumeration of the KMT2A fusion interactome, and the diff against this graph

  counter-evidence repair: the WRAD discrepancy is a co-IP negative about stable complex membership, not evidence that WRAD is irrelevant to KMT2A-rearranged leukemia

• ← related toMLLT6 (AF17) — the fourth KMT2A fusion partner inside DotCom, and the only one no cell needs

  methodological: DotCom membership rests on unquantified MudPIT co-purification; quantitative stoichiometry reassigned DPY30's complex membership when it was applied to SET1/MLL

• ← bindsDPY30 D/D domain and the ASH2L SDI helix: four hydrophobic pockets, one load-bearing leucine - and a 94-residue numbering offset that relocates the whole motif

  D/D domain 45-99 as an obligate dimer

• ← related to18% of this graph's evidence is the same finding said again

  One of the four WRAD subunit entries carrying identical copies of complex-level evidence.

• ← refutesIs the selectivity just pan-essentiality? No: it explains 0.4% of the variance

  Does NOT rescue DPY30. Adjusted for essentiality its residual z is -0.76, inside the noise, consistent with its failure on the within-subtype, line-normalised and variance-matched designs.

• ← refutesThe assumption the whole project rests on, tested at last: menin dependency is fusion-specific (run)

  DPY30 fails a fourth independent design, at the 14th percentile of random genes

• ← related to49 of 91 proteins in this graph have never had a drug-like molecule bound to them

  Full construct coverage across 12 entries and zero ligands - independent of the DepMap designs on which DPY30 already fails.

• ← refutes4,589 genes with no prior, and menin comes first (run)

  unremarkable at the 14.7th percentile, the fifth independent design to say so

• ← related toDPY30 has a second job at the trans-Golgi network, and the authors who found it could not put ASH2L or RBBP5 there

  DPY30 has an experimentally coded trans-Golgi network pool as well as a nuclear one

• ← bindsKMT2A-AKAP8 (AKAP95): graded 'prey-only' by this project, and it is a direct positive modulator of MLL H3K4 methyltransferase activity from a dedicated MLL paper - ADMITTED

  direct: purified AKAP95 residues 1-100 bind purified DPY30

• ← bindsAKAP8 (AKAP95) - the only nuclear A-kinase anchoring protein, and the one that turned out to be required for MLL-rearranged leukaemia

  AKAP95 N-terminus (1-210) to the DPY30 Dpy-30 motif; the region assignment is disputed between two groups

• ← bindsAKAP95-DPY30: a direct contact whose binding determinant two groups map to OPPOSITE ends of AKAP95, and neither has been structurally resolved

  direct, bacterially purified on both sides; DPY30 side maps to the Dpy-30 dimerisation motif and L69D abolishes it

• ← related toAKAP95 is phenotypically silent at steady state and required only when a cell is changing state - the same shape in ATRA differentiation and in MLL-AF9 transformation, thirteen years apart

  AKAP95 and DPY30 depletion affect overlapping ATRA target sets, but AKAP95's own targets are enriched for ECM and signalling rather than transcriptional regulators

• ← refutesThe leukaemia term hid the answer for DPY30 — a rule-77 demonstration with clean rule-67 controls

  the best cell-state result for DPY30 is not the murine adult-HSC one: knockdown in HUMAN CD34+ progenitors impairs myelomonocytic differentiation while POTENTLY PROMOTING haemoglobin production, with severe haematopoietic defects in zebrafish and a crucial role in MLL1-fusion leukaemia line growth (PMID 25139354) - which is content the leukaemia-scoped query that returned zero was looking for

• ← related toTwo thirds of the curated interactome of this project's six core proteins is spoke-expanded - and the pairs this graph chose to write are significantly better than the ones it did not

  DPY30-KMT2A is 3 of 3 spoke-expanded and shares PDB 6PWV, where DPY30 is present through ASH2L and does not contact KMT2A - flag and structure agree

• ← related toLost from the KMT2A half: the WIN motif and the whole WRAD interface (aa 3762-3767, Arg3765)

  WRAD subunit that cannot assemble on the fusion

• ← refutesThe founding WRAD negative was never run on a fusion protein

  Scope error: Yokoyama 2004 does not mention DPY30 anywhere (zero occurrences of 'DPY' in the full text) and DPY30 was not placed in the human SET1-like complexes until 2007, so E_yokoyama2004 cannot carry a negative about DPY30.

• ← related toDPY30: the founding negative does not cover it, and it has no KMT2A contact to lose

  Splits this subunit's evidence into what is established for wild-type KMT2A and what, if anything, is established for a fusion.

• ← refutesDPY30: the founding negative does not cover it, and it has no KMT2A contact to lose

  DPY30 has no mapped KMT2A contact at all; its interface is with the ASH2L SDI motif, so the retained/lost framing reaches it only through ASH2L.