ZNF512 (Q96ME7) - the one protein of the nine deferred pan-MLL-fusion preys that is a sequence-specific chromatin factor, and it initiates pericentric heterochromatin
also called ZNF512 · KIAA1805 · Zinc finger protein 512 · ZN512_HUMAN · Q96ME7
ZNF512 is a DNA-binding protein that parks itself on the repetitive DNA near the centre of chromosomes and calls in the enzymes that shut that DNA down. Of the nine proteins in this batch it is the only one whose day job is on chromatin, which is where an MLL fusion lives - so it is the one most likely to be a genuine partner rather than a laboratory accident.
The same thing, for a scientist
ZNF512 (Q96ME7, 567 aa) is a sequence-specific zinc-finger protein that, with its paralogue ZNF512B, binds pericentric repeats directly and is SUFFICIENT to initiate de novo heterochromatin by recruiting SUV39H1 and SUV39H2 to catalyse H3K9 methylation; the atypical long linkers between its zinc fingers let it read non-consecutively organised triplets, which is why the same machinery targets pericentric sequence that is not conserved between vertebrates (Nature 2024). It was independently identified at mouse chromocentres by dCas9-BirA* proximity labelling on major satellite DNA and confirmed by super-resolution microscopy. In Skucha 2018 it co-purified with all seven MLL fusions (MLL-AF4 p = 0.0463, 4|6, negative control 0|2, rank 34 of 42). **The module measurement made here is the reason this entry is the strongest of the nine:** TRIM28/KAP1 - the canonical heterochromatin corepressor scaffold - is ALSO pan-fusion (7 of 7), already has an entry in this graph, and sits beside ZNF512 in the same column. Against that: SUV39H1, SUV39H2, CBX1, CBX3, CBX5, ATRX, DAXX and ZNF512B are all absent from the 960, so the constitutive-heterochromatin module is two proteins deep and no more. UniProt's nuclear and pericentric-heterochromatin calls for human ZNF512 are ECO:0000250 by similarity to mouse Q69Z99 and are NOT human experimental - the human-relevant evidence is Ma 2024, not the database row.
The Skucha row, stated in full
7 of 7 fusions. MLL-AF1p 0.00905 (6|6), MLL-AF4 0.0463 (4|6), MLL-AF9 0.000573 (15|20), MLL-CBP 0.00556 (5|22), MLL-EEN 0.00107 (23|20), MLL-ENL 9.24e-05 (23|22), MLL-GAS7 0.00379 (16|17); negative control 0|2 1. Rank 34 of 42 by MLL-AF4 p. Six of the seven baits are below 0.01; only the index fusion's column is marginal, and its spectral total of 10 is rank 7 of the seven baits against 45 under MLL-ENL. The non-zero control (0|2) is one of only ten in the 42 and, on this beat's abundance measurement, sits exactly where a prey reaching 45 spectra would be expected to.
What ZNF512 actually does, which is chromatin
ZNF512 and ZNF512B localise to pericentric regions through direct DNA binding, and both are sufficient to initiate de novo heterochromatin formation at ectopically targeted repetitive regions and at pericentric regions, by directly recruiting SUV39H1 and SUV39H2 to catalyse H3K9 methylation. The mechanism that makes this work across species is structural: the atypical long linker residues between the zinc fingers give flexibility to recognise non-consecutively organised three-nucleotide triplets, so ZNF512 proteins from one vertebrate can target another's pericentric regions despite no sequence conservation 2. Independently, dCas9-BirA* proximity labelling targeted to major satellite DNA recovered ZNF512 among the local chromatin proteome, and super-resolution microscopy placed it at chromocentres 3.
This is the only protein of the nine whose normal address is chromatin. MAP7D1, MAPRE1 and LUZP1 are cytoskeletal, VIM and ACTB are structural, EIF4G1 is translational, PRKAR1A is a kinase regulatory subunit annotated only at the cell membrane, and DHX16 is nucleoplasmic but spliceosomal. ZNF512 reads DNA.
The module: TRIM28 is pan-fusion too, and the rest of the machinery is not
Measured against the full 960-row table on 2026-09-21. Present: TRIM28/KAP1 at 7 of 7 fusions (MLL-AF4 p = 0.129, neg 0|0) - the canonical corepressor scaffold of KRAB-zinc-finger heterochromatin - which already has an entry in this graph. HP1BP3 at n = 2 with a dirty 4|4 control. Absent at any n: SUV39H1, SUV39H2 (the enzymes ZNF512 recruits), CBX1, CBX3 and CBX5 (all three HP1 paralogues), ATRX, DAXX, SMCHD1, and ZNF512B, ZNF512's own obligate-partner paralogue.
So the constitutive-heterochromatin signal in this network is two proteins - ZNF512 and TRIM28 - and neither of the enzymes nor any HP1 came with them. That is much less than a module and more than nothing. It is worth stating in the direction that is testable: if MLL fusions were touching pericentric heterochromatin, SUV39H1 and an HP1 should be in the 960, and they are not 12.
The database row is by similarity; the human evidence is the paper
UniProt Q96ME7, fetched live 2026-09-21 with per-location evidence codes read individually: 'Nucleus' and 'Chromosome' are both ECO:0000250 by similarity to mouse Q69Z99, with the uncoded note 'Associated with pericentric heterochromatin'. There is no human experimental localisation annotation at all. Per rule 91 the uncoded note was read rather than discarded, and per the same rule the pointer was checked: the founding human-relevant observation is Ma 2024, which UniProt does not cite for these fields. This entry records the compartment on the paper, not the database.
Verdict
The most likely real partner of the nine, and still only weak. It is a sequence-specific chromatin protein with a first-rate mechanism paper, it recurs across all seven baits with six of seven below p = 0.01, and the one other constitutive-heterochromatin protein in the pan-fusion set already has an entry here. What holds the confidence down is that the index fusion's own column is the marginal one (p = 0.0463, rank 34 of 42), that neither SUV39H nor any HP1 nor ZNF512B accompanies it, and that no experiment outside this AP-MS has ever put ZNF512 near a KMT2A fusion - the bounded query returns 0.
| uniprot | Q96ME7 |
|---|---|
| length AA | 567 |
| accession resolution | Resolved live 2026-09-21: `gene_exact:ZNF512 AND organism_id:9606 AND reviewed:true` returns exactly ONE reviewed row; gene_primary ZNF512, Q96ME7, entry name ZN512_HUMAN, 567 aa, protein name 'Zinc finger protein 512', synonym KIAA1805. Matches the accession in Skucha's own table. Note the paralogue ZNF512B (Q96KM6) is a DIFFERENT gene with a different accession and is ABSENT from the Skucha 960 - a swap would be easy and is guarded here by both accession and length. |
| identifier hazard | NOT IN data/symbol-collisions.json AS OF 2026-09-21 (registry read before searching, per rule 88). Two live hazards. (1) `Circ_ZNF512` is a circular RNA with its own cardiology literature (PMID 35041407), so a bare ZNF512 search collects records about a transcript that is not this protein - 1 of 11 records on the bare query. (2) ZNF512 and ZNF512B are functionally interchangeable in the founding paper and are constantly written together, so a paper 'about ZNF512' is often about both; ZNF512B is absent from the 960 and the two must not be merged. Rule 65 readback SUCCEEDS. |
| localization | NUCLEUS AND CHROMOSOME, BY SIMILARITY IN THE DATABASE AND EXPERIMENTALLY IN THE LITERATURE - and those are different sources. UniProt Q96ME7, fetched live 2026-09-21 with per-location evidence codes read individually: 'Chromosome' ECO:0000250 from mouse Q69Z99; 'Nucleus' ECO:0000250 from mouse Q69Z99; uncoded note 'Associated with pericentric heterochromatin'. There is NO human experimental localisation annotation. The compartment is nevertheless well established in the literature: direct pericentric DNA binding and de novo heterochromatin initiation (Ma 2024), and super-resolution imaging at mouse chromocentres after CasID proximity labelling on major satellite DNA (Schmidtmann 2016). Recorded on the papers, not the database row, per rule 91. |
| subnuclear localization | PERICENTRIC HETEROCHROMATIN / CHROMOCENTRES. This is the only member of this tranche with a positive subnuclear assignment. Sources: Ma 2024 (direct DNA binding at pericentric regions, sufficiency for de novo heterochromatin) and Schmidtmann 2016 (super-resolution localisation at chromocentres in mouse cells). Both are non-haematopoietic and the imaging is murine. NOT measured in any human leukaemia cell and NOT measured in any KMT2A-rearranged cell. |
| localization shown for fusion | NO - nobody has imaged or fractionated ZNF512 in any cell expressing any KMT2A fusion. This graph holds `localization-chromocentres-pericentromeric-kmt2a-r`, which frames the pericentromeric question for KMT2A-rearranged cells; ZNF512 is the first pan-fusion interactor with a documented chromocentre address to link to it. |
| shown for fusion | Co-purification with all seven MLL fusions including MLL-AF4 in Jurkat (p = 0.0463, six of seven baits below 0.01); shRNA-screened in MLL-AF9 MOLM-13, score unpublished. NO independent experiment ties ZNF512 to any KMT2A fusion in any system. |
| selectivity | UNKNOWN. Not named among the paper's controls (RFC1, SF3A3, CCT3 non-selective; MEN1 selective) and not known to be among the 40 re-screened, a list the paper does not publish. Absence from the named controls is NOT evidence of selectivity. NOT barred from future consideration the way the cytoskeletal members of this tranche are - a sequence-specific zinc-finger protein is a more plausible object of study than beta-actin - but this entry does not nominate it, because no selectivity measurement of any kind exists and the index-fusion call is the weakest of its seven. |
| absence class | never-measured |
| absence class note | No instrument has been pointed at ZNF512 in a KMT2A-rearranged cell outside the two screen hosts. The MOLM-13 shRNA score is measured-not-reported. PRIDE checked: PXD009338 is the only relevant deposition. Note a second, cheaper absence class hiding here: whether SUV39H1/2 or any HP1 appears in PXD009338 below the 300-per-bait rank cutoff is measured-not-analysed - the spectra are public. |
| graph context | ZNF512 appears in ZERO of this graph's 1,924 node files before this entry. TRIM28, the other pan-fusion heterochromatin protein, does have an entry (`gene-trim28`, confidence weak, MLL-AF4 p 0.129). |
| verdict | THE MOST LIKELY REAL PARTNER OF THE NINE, held at `weak`. It is the only one of the nine whose normal address is chromatin; it has a first-rate mechanism paper; six of seven baits are below p = 0.01; and the one other constitutive-heterochromatin protein in the pan-fusion 42 is already written here. Against: the index-fusion column is the marginal one, no SUV39H, no HP1 and no ZNF512B accompany it, and no independent experiment has ever put it near a KMT2A fusion. |
skucha 2018
{
"bait": "MLL-AF4 (= KMT2A::AFF1)",
"p_value": 0.0463,
"spectral_counts": "4|6",
"n_fusions_of_seven": 7,
"negative_control_counts": "0|2",
"per_bait": {
"MLL-AF1p": "0.00905 (6|6)",
"MLL-AF4": "0.0463 (4|6)",
"MLL-AF9": "0.000573 (15|20)",
"MLL-CBP": "0.00556 (5|22)",
"MLL-EEN": "0.00107 (23|20)",
"MLL-ENL": "9.24e-05 (23|22)",
"MLL-GAS7": "0.00379 (16|17)"
},
"af4_spectral_rank_of_seven": 7,
"af4_p_rank_within_the_42_pan_fusion": 34,
"index_bait_note": "Six of seven baits at or below 0.00905. The index fusion's column is the ONLY marginal one, which on this beat's bait-depth measurement is the expected shape and not a statement about ZNF512.",
"source": "analyses/skucha-mll-fusion-interactome/results.json, re-derived from the cached MOESM4 xlsx 2026-09-21; run.py re-run, both controls PASS"
}complex membership
{
"complex": "constitutive/pericentric heterochromatin initiation (ZNF512 + ZNF512B -> SUV39H1/2 -> H3K9me)",
"partners_in_the_960": {
"TRIM28": "n=7, MLL-AF4 p 0.129 (neg 0|0) - the KRAB-ZNF corepressor scaffold; already has an entry in this graph",
"HP1BP3": "n=2, MLL-AF4 p 0.0624, neg 4|4 - a dirty control and only two baits"
},
"absent_from_the_960": [
"ZNF512B",
"SUV39H1",
"SUV39H2",
"CBX1 (HP1beta)",
"CBX3 (HP1gamma)",
"CBX5 (HP1alpha)",
"ATRX",
"DAXX",
"SMCHD1",
"SETDB1"
],
"reading": "TWO PROTEINS DEEP AND NO MORE, and the absences are the informative part. Neither enzyme ZNF512 recruits, no HP1 paralogue, and not even ZNF512B's own paralogue is present. No heterochromatin module was tested in analyses/mll-fusion-interactome-complexes/, so this pairing has NOT been tested against a background and is an observation, not an enrichment claim (rule 84)."
}cell types and states
[ "Jurkat (human T-ALL) - the AP-MS host, nuclear lysate, tagged single-copy transgene", "MOLM-13 (human MLL-AF9 AML) - shRNA screen host, six shRNAs, per-gene score unpublished", "mouse and other vertebrate cells - the pericentric heterochromatin mechanism and the chromocentre imaging", "no human leukaemia cell outside the two screen hosts, no B-lineage cell, no infant cell, and no KMT2A::AFF1 cell" ]
functionally screened in the source paper
{
"screened": true,
"shrnas_for_this_gene": 6,
"design": "Supplementary Table 6 / MOESM5 is the shRNA library for the negative-selection screen: 768 data rows, 128 unique gene symbols, EXACTLY 6 shRNAs each - counted with openpyxl by this beat on 2026-09-21. Host is MLL-AF9-expressing MOLM-13, a human myeloid line.",
"set_identity_measured_by_this_beat": "The 128 symbols in the shRNA library are the SAME SET as the 128 rows of MOESM4 with n_fusions >= 5, checked by set difference on 2026-09-21: the only four mismatches are alias spellings (JADE2/PHF15, JADE3/PHF16, KMT2A/MLL, NIFK/MKI67IP). This matters for the contamination question. The paper says 'Seven proteins were manually removed as these were either frequently observed contaminants or were not detected in human hematopoietic cell lines. The final set was comprised of 128 proteins.' Because the table's >=5 set is already exactly 128, those seven are NOT recoverable from Supplementary Table 5 - and every protein in this tranche is in the post-removal 128 and was carried into the functional screen.",
"what_this_does_NOT_establish": "A per-gene result. Scores are presented only as heatmaps (Fig. 2c/2d); no per-gene depletion value is tabulated anywhere in the paper or its supplementary files, so for any individual gene the outcome is measured-not-reported. It is also MLL-AF9 MOLM-13, a myeloid line, not the index fusion and not B-lineage.",
"the_filter_is_leaky_and_here_is_the_proof": "'Survived the authors' manual contaminant removal' is weaker than it sounds. CCT3 is also in the 128 and was also screened, and the CCT/TRiC chaperonin is the canonical AP-MS contaminant - the companion analysis analyses/mll-fusion-interactome-complexes/ uses it as its NEGATIVE control and it is still significant against a nuclear background (q = 2.7e-05, 8 of 9 subunits in the MLL-AF4 column). The authors' own named essentiality controls were RFC1, SF3A3 and CCT3. So the manual removal did not remove a canonical contaminant complex, and membership in the 128 is not the authors certifying a protein as a real partner.",
"absence_class_for_the_per_gene_score": "measured-not-reported",
"library_symbol": "ZNF512"
}bounded literature
{
"query": "ZNF512[tiab] AND (KMT2A[tiab] OR MLL[tiab] OR \"MLL-AF4\"[tiab] OR \"MLL-AF9\"[tiab] OR \"t(4;11)\"[tiab])",
"querytranslation": "\"ZNF512\"[Title/Abstract] AND (\"KMT2A\"[Title/Abstract] OR \"MLL\"[Title/Abstract] OR \"MLL-AF4\"[Title/Abstract] OR \"MLL-AF9\"[Title/Abstract] OR \"t 4 11\"[Title/Abstract])",
"count": 0,
"date": "2026-09-21",
"tool": "NCBI esearch db=pubmed",
"names_tried": [
"ZNF512",
"KIAA1805"
],
"bare_operand_control": "ZNF512[tiab] alone returns 11 records, so the gene clause can return ones.",
"positive_control": "MEN1[tiab] AND the identical five-term KMT2A clause returns 85.",
"bound": "Titles and abstracts only; supporting figures not searched (rule 80). Cytogenetic term quoted, querytranslation read back with no injected [UID] term (rule 95). One tag per term (rule 94). The paralogue form ZNF512B was NOT run against the KMT2A clause and is a residual gap."
}what would settle it
[ "Search PXD009338 for SUV39H1, SUV39H2, CBX1/3/5 and ZNF512B peptides below the 300-per-bait rank cutoff. Public data, no wet work, and it decides whether this is a truncated heterochromatin module or an isolated hit.", "ChIP ZNF512 in a KMT2A::AFF1 B-ALL line and ask whether it is anywhere other than pericentric repeats. If it is at MLL-fusion target promoters, this entry should be upgraded.", "Reciprocal co-IP of ZNF512 against the fusion in SEM or RS4;11. At 10 spectra in the index column this is the honest test." ]
- ZNF512 must not be nominated as a target from this entry. There is no selectivity measurement of any kind, the index-fusion call is the weakest of its seven baits, and initiating constitutive heterochromatin is a housekeeping function.
- The specific trap here is that a good mechanism paper makes a weak AP-MS row feel strong. The Nature 2024 work is excellent and is about pericentric repeats in vertebrate cells; it supplies no evidence whatever about MLL fusions, and this entry's confidence is set by the AP-MS row, not by the paper's quality.
- If ZNF512 did prove to be a genuine fusion partner, the therapeutic read would be discouraging: disrupting de novo H3K9 methylation initiation is a global chromatin intervention with an obvious genome-instability cost, in a disease whose own literature already links SETD2 loss to DNA damage.
- Are SUV39H1/2 or any HP1 paralogue present in PXD009338 below the rank cutoff? The whole heterochromatin reading of this row turns on it and the data are public.
- Does ZNF512 bind anywhere other than pericentric repeats in a leukaemia cell? Its function was unknown until 2024 and no human haematopoietic ChIP exists.
- TRIM28 and ZNF512 are both pan-fusion. Are MLL fusions near constitutive heterochromatin at all? This graph holds `localization-chromocentres-pericentromeric-kmt2a-r` as an open question and these are the first two interactome proteins that speak to it.
- ZNF512B is functionally equivalent to ZNF512 in the founding paper and is completely absent from the 960. Is that a real discrimination by the bait, or a detection floor? A paralogue-resolved reanalysis of the spectra would say.
- E1weak confidence 2 of 5: Preliminary, small, indirect, or not yet peer reviewed.paper
ZNF512 co-purified with all seven Strep-HA-tagged MLL fusions from nuclear lysates of isogenic human Jurkat T-ALL cells; for MLL-AF4 (= KMT2A::AFF1) the combined p-value is 0.0463 with spectral counts 4|6 and a negative control of 0|2, six of seven baits are below p = 0.01, and it is in the paper's final post-filter set of 128 that entered the MOLM-13 shRNA screen.
MLL-fusion-driven leukemia requires SETD2 to safeguard genomic integrity. · Skucha A, Ebner J, Schmoellerl J, Roth M, Eder T, Cesar-Razquin A, Stukalov A, Vittori S, Muhar M, Lu B, Aichinger M, Jude J, Mueller AC, Gyorffy B, Vakoc CR, Valent P, Bennett KL, Zuber J, Superti-Furga G, Grebien F · Nature Communications 2018What this evidence cannot show
SYSTEM: the AP-MS host is JURKAT, a human T-ALL line, while KMT2A::AFF1 disease is B-lineage infant leukaemia; a partner requiring a B-cell or infant proteome cannot appear here and a T-lineage partner can appear spuriously. The bait is a Strep-HA-tagged, SINGLE-COPY, doxycycline-induced transgene at an FRT/Flp cassette, not the endogenous fusion. Prey came from bulk nuclear lysate after a single STREP-Tactin step, so co-purification is NOT evidence of a direct contact and is not evidence that the two proteins are ever in the same complex. INCLUSION RULE: the paper kept a fixed number of top-ranked interactions per bait rather than applying a p threshold. Methods, read verbatim from the PMC full text by this beat on 2026-09-21: "All the identified bait-prey pairs were ranked by the combined p-value. The 300 most significant interactions per bait were retained." Measured by this beat from the cached MOESM4 xlsx: within the MLL-AF4 column 186 of 291 listed p-values (63.9%) exceed 0.05, and that column's maximum p is 0.175. BAIT DEPTH: for all nine proteins in this tranche the MLL-AF4 spectral total is rank 7 of the seven baits - measured protein by protein - so a low MLL-AF4 count is a bait property (transgene expression, purification depth, or the GLM normalisation) and must not be read as a weaker association than the same protein's count under another bait. HOW TO READ THE NEGATIVE CONTROL, measured by this beat on 2026-09-21 over all 960 rows of MOESM4: the Spearman correlation between a protein's neg_ctrl spectral total and its maximum per-bait spectral total is +0.461, and the median maximum per-bait spectral count is 42 for the 145 proteins (15.1%) with a non-zero negative control against 8 for the 815 with 0|0. A dirty negative control in this table is therefore substantially a statement about protein ABUNDANCE, not about specificity, and a clean 0|0 is substantially a statement that two control replicates were not deep enough to see the protein at all. Two consequences. (1) PSIP1/LEDGF - the one pan-fusion partner with a solved co-crystal against the KMT2A N-terminus - carries 0|5, the second-dirtiest control of the 42, so a non-zero control cannot be used to call a protein a contaminant. (2) A 0|0 control is not purity; it is an unestimated baseline. The paper's GLM estimates each prey's baseline abundance A0,j from the control purifications, so a prey the controls never saw has its baseline set by a floor rather than by data, which is the direction that inflates apparent specificity. WHERE THIS ROW SITS AMONG THE 42, measured by this beat on 2026-09-21: ranked by MLL-AF4 p-value the pan-fusion set runs from MEN1 at 9.75e-16 to SMC3 at 0.174. The three WEAKEST MLL-AF4 calls in the whole pan-fusion set - SMC3 0.174, NOLC1 0.134, TRIM28 0.129 - all already have entries in this graph, and gene-nolc1 and gene-trim28 both carry confidence_overall 'weak' with those p-values recorded. So p > 0.05 in the index-fusion column is not a disqualifier in this corpus; it is a confidence tier that two existing entries already occupy. SPECIFIC TO THIS PROTEIN, measured by this beat from the cached MOESM4 xlsx on 2026-09-21: MLL-AF4 p = 0.0463 is rank 34 of 42 and barely clears 0.05, while six of the seven baits are at or below 0.00905 - so the pan-fusion status is solid and the INDEX-fusion call is the weak one. Its 0|2 negative control is one of only ten non-zero controls in the 42 and is unremarkable given a maximum per-bait spectral total of 45. MODULE: TRIM28/KAP1 is also 7 of 7 (MLL-AF4 p 0.129, neg 0|0) and already has an entry here, but SUV39H1, SUV39H2, CBX1, CBX3, CBX5, ATRX, DAXX, SMCHD1 and ZNF512B are ALL absent from the 960 at any n. SELECTIVITY: not named among the paper's controls (RFC1, SF3A3, CCT3 non-selective; MEN1 selective), not known to be among the unpublished 40 re-screened. Recorded as unknown. run.py re-run 2026-09-21, both controls PASS.
Evidence that argues against this (2)
- If MLL fusions genuinely engaged pericentric heterochromatin, the enzymes and readers should be in the network with ZNF512 and they are not - no SUV39H1, no SUV39H2, no HP1 paralogue, no ATRX, no DAXX, and not even ZNF512B, the paralogue that does the same job at the same sites. The recovered constitutive-heterochromatin signal is two proteins deep. IntAct has never curated this dataset; the raw spectra at PXD009338 have not been reanalysed here, and a search of them for SUV39H and CBX peptides below the rank cutoff is the decisive missing check. ProteomeXchange/PRIDE 2018
- This project's own complex analysis, once corrected to a realistic 5,754-protein nuclear background rather than the 20,431-protein proteome, finds seven of twelve complexes stop being significant - only Nucleolar/ribosome biogenesis, NEXT/exosome and HBO1/KAT7 survive alongside the two controls. No heterochromatin module was tested at all, so no module claim is available for ZNF512 from that analysis, and the two-protein ZNF512/TRIM28 observation in this entry has NOT been tested against a background and must not be read as an enrichment (rule 84). internal analysis 2026
- E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
ZNF512 and ZNF512B localise specifically at pericentric regions through direct DNA binding, and both are sufficient to initiate de novo heterochromatin formation at ectopically targeted repetitive regions and at pericentric regions by directly recruiting SUV39H1 and SUV39H2 to catalyse H3K9 methylation; atypical long linker residues between their zinc fingers allow recognition of non-consecutively organised nucleotide triplets, so ZNF512 and ZNF512B from one vertebrate species can target the pericentric regions of another despite the absence of DNA sequence conservation.
Targeting pericentric non-consecutive motifs for heterochromatin initiation. · Ma R, Zhang Y, Zhang J, Zhang P, Liu Z, Fan Y, Wang HT, Zhang Z, Zhu B · Nature 2024What this evidence cannot show
Vertebrate cell systems with ectopic tethering as the sufficiency assay; the pericentric biology is worked out across species rather than in a human haematopoietic cell, and nothing in it concerns leukaemia, KMT2A or any fusion. SUV39H2 contributes more to H3K9 trimethylation while SUV39H1 contributes more to silencing, so the two arms of the recruitment are not equivalent. Read from the abstract. This is the human-relevant source for ZNF512's compartment and function; UniProt's own nuclear and chromosome annotations for Q96ME7 are ECO:0000250 by similarity to mouse Q69Z99 and do not cite this paper.
Evidence that argues against this (1)
- This establishes what ZNF512 does at pericentric heterochromatin and says nothing about what it might do at an MLL fusion's targets, which are CpG-island promoters and gene bodies - close to the opposite chromatin environment. A protein can be a first-rate chromatin factor and still be an AP-MS artefact, and the pan-fusion network contains no SUV39H, no HP1 and no ZNF512B to corroborate a pericentric connection. Skucha A et al. 2018
- E3moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper
dCas9 fused to the promiscuous biotin ligase BirA* (CasID), targeted to repetitive telomeric, major satellite and minor satellite DNA, biotinylated and recovered both known and previously unknown chromatin factors by streptavidin precipitation and mass spectrometry; super-resolution microscopy confirmed the localisation of the putative transcription factor ZNF512 at chromocentres.
Determination of local chromatin composition by CasID. · Schmidtmann E, Anton T, Rombaut P, Herzog F, Leonhardt H · Nucleus 2016What this evidence cannot show
MOUSE cells - major and minor satellite are murine pericentric and centromeric repeats - and the paper is a method report, with ZNF512 as the one hit taken forward to imaging. Proximity labelling reports a neighbourhood, not a contact. Not haematopoietic, not human, not a fusion system. Read from the abstract. Included because it is an independent route to the same compartment eight years before the mechanism paper, and because it is the kind of experiment - locus-directed proximity labelling - that this graph records as never having been done on a KMT2A fusion.
Evidence that argues against this (1)
- A proximity-labelling hit at mouse chromocentres does not transfer to a human T-ALL nuclear lysate, and the source paper itself frames ZNF512 as a 'putative' transcription factor - its function was unknown until 2024. This citation corroborates the compartment, not the interaction, and the compartment it corroborates is not where MLL fusions work. Ma R et al. 2024
co-purified, all 7 MLL fusions, Jurkat nuclear-lysate AP-MS, MLL-AF4 p=0.0463 (4|6), neg 0|2, six of seven baits below 0.01 - the strongest candidate of the nine
the only two constitutive-heterochromatin proteins in the pan-fusion 42, both 7 of 7 - and SUV39H1/2, all three HP1 paralogues, ATRX, DAXX and ZNF512B are absent, so the module is two deep
first pan-fusion interactor with a documented chromocentre address: direct pericentric DNA binding and de novo H3K9 methylation via SUV39H1/2 (Ma 2024), chromocentre imaging by CasID (Schmidtmann 2016)
- related to →KMT2A / MLL1 — the gene that breaks
pan-MLL-fusion core prey; the shared bait is the KMT2A N-terminus