PSIP1-FASTKD2 declined: the bait is a mitochondrial matrix and nucleoid protein with two experimental localisations, the prey is nucleoplasmic LEDGF, and the row is one spoke-expanded BioID2 proximity in K562
also called LEDGF-FASTKD2 proximity row · FASTKD2 bait PSIP1 prey · IM-30059 FASTKD2 PSIP1
FASTKD2 lives inside mitochondria, where it helps build the mitochondrial ribosome. LEDGF lives in the nucleus on chromatin. A proximity screen put them together in one row, and we are declining it, because two experiments place FASTKD2 in a compartment that a nuclear protein cannot be in.
The same thing, for a scientist
**Declined on compartment.** FASTKD2 (HGNC:29160, previous symbol **KIAA0971**, ENSG00000118246, UniProt **Q9NYY8**, 710 aa, 2q33.3) is the *FASTK mitochondrial RNA regulator 2*. UniProt assigns it **mitochondrion matrix and mitochondrion nucleoid with ECO:0000269 experimental evidence from two independent papers** (PMID 20869947, PMID 25683715), plus a third ECO:0000305 matrix call; its only annotated domain is a **RAP domain at 634-691**; and HPA's immunofluorescence main location is **Mitochondria**, sole, at the *Supported* tier [E3][E4]. PSIP1/LEDGF is nucleoplasmic. The row is IntAct **IM-30059**, **MI:1314 BioID2**, **MI:1060 spoke**, **FASTKD2 bait / PSIP1 prey**, MI-score 0.27, K562 [E1]; FASTKD2 recovered 237 of 2,681 preys (33rd of 50, panel median 326) and PSIP1 is recovered by 9 of the 50 baits, giving **p = 0.113** under a both-margins null - NOT RESOLVED by the statistic [E2]. **The decline is not statistical.** A mitochondrial-matrix BioID2 bait biotinylates the matrix; a nucleoplasmic prey in its list is either a sub-stoichiometric extra-mitochondrial pool, a post-lysis association, or a streptavidin-capture background, and none of those is an interaction. **PubMed returns zero for FASTKD2 against KMT2A, zero against MLL, and zero against PSIP1 or LEDGF** [E5].
No three-dimensional model is available for this protein, so the map above is the whole picture. That is itself informative: AlphaFold does not model chains this long, and the parts that have been crystallised are held under the entries for the individual domains.
The compartment argument, which is the whole decline
| source | call | tier |
|---|---|---|
| UniProt Q9NYY8 | Mitochondrion matrix, mitochondrion nucleoid | ECO:0000269, PMID 20869947 and PMID 25683715 - two independent experimental papers |
| UniProt Q9NYY8 | Mitochondrion matrix | ECO:0000305, PMID 18771761 - curator inference |
| HPA v25.1 IF | Mitochondria, main and only | Supported |
| HPA function | rRNA-binding, ribosome biogenesis | gene-level |
This is the best-evidenced localisation of any protein in this beat, and it is unambiguous. Simarro et al. confirmed the mitochondrial localisation of FASTKD2 and showed that all FASTKD family members are found in mitochondria 3. Antonicka and Shoubridge characterised the mitochondrial RNA granule proteome and showed FASTKD2 is required for mitochondrial ribosome biogenesis and binds the 16S rRNA 4. FASTKD2 mutations cause a mitochondrial encephalomyopathy with reduced cytochrome c oxidase activity.
PSIP1/LEDGF is a nucleoplasmic chromatin protein - PWWP domain reading H3K36me2/3 on nucleosomes, IBD engaging the MLL IBM, HPA main location nucleoplasm at the Enhanced tier. There is no shared compartment.
This graph has already declined four PSIP1 partners on precisely this ground - interface-psip1-gdf5-declined-a-secreted-cartilage-growth-factor-crosslinked-in-nuclei, interface-psip1-ighmbp2-declined-compartment-compatible-single-crosslink, interface-psip1-ppp1r13b-declined-aspp1-single-crosslink-in-unfractionated-nuclei and interface-psip1-tmem214-declined-an-er-membrane-protein-crosslinked-in-nuclei - and one KMT2A partner, interface-kmt2a-tmco1-declined-an-er-calcium-channel-crosslinked-in-nuclei. Those five come from a crosslinking study; this is the same verdict reached for the first time on a BioID2 row, which is a different method with the same failure mode.
The two things that make this a decline and not a dismissal
A mitochondrial protein CAN have an extra-mitochondrial pool, and the honest version of this decline says so. What the row would require is that a fraction of FASTKD2 - fused to BioID2, over-expressed, one isoform - resides long enough near nucleoplasmic LEDGF to biotinylate it. Nothing in the two localisation papers looked for such a pool and nothing excludes one. The decline is that no evidence supports it, not that it is impossible.
And the existing co-dependency signal is weaker than it looks. ana-depmap-co-dependency-cannot-rank-the-interactome-tail-either-and-my-pass-criterion-was-a-count-without-a-baseline records a FASTKD2-PSIP1 DepMap co-dependency of r = 0.125 over 1,208 shared lines. That entry's whole point is that its pass criterion was a count without a baseline, and r = 0.125 over 1,208 lines is a correlation that is statistically distinguishable from zero and biologically indistinguishable from nothing - it explains 1.6% of the variance. It is recorded here so the next beat does not find it and read it as corroboration.
The row, the literature, and where this is filed
IMEx IM-30059: MI:1314 proximity-dependent biotin identification (BioID2), MI:1060 spoke expansion, roles MI:0496 bait (FASTKD2) / MI:0498 prey (PSIP1), host taxid:9606(human-k_562), MI-score 0.27 1. FASTKD2 is one of the 50 RBP baits; the authors themselves state that RBP over-expression, BioID2 fusion and single-isoform use may bias the data.
Bounded absence, both operands 5:
| query | translation | n |
|---|---|---|
FASTKD2[tiab] AND KMT2A[tiab] | "FASTKD2"[Title/Abstract] AND "KMT2A"[Title/Abstract] | 0 |
FASTKD2[tiab] AND MLL[tiab] | "FASTKD2"[Title/Abstract] AND "MLL"[Title/Abstract] | 0 |
FASTKD2[tiab] AND (PSIP1[tiab] OR LEDGF[tiab]) | non-empty | 0 |
FASTKD2[tiab] whole literature | - | 43 |
All genuine zeros - non-empty translations, warninglist reading No items found. HGNC's previous symbol KIAA0971 was not queried and is the spelling in pre-2003 sequencing literature.
Filed as an interface-*-declined-* node deliberately. A gap list matching own_entry against gene-* and protein-* ids cannot see this node, which is the failure this beat measured at 2 of 22 in its own class - so the symbol is in the title and the aliases, where a grep will find it.
| partner a | gene-psip1 |
|---|---|
| partner b | FASTKD2 (no entry - declined, this node is the record) |
| verdict | DECLINED |
| verdict basis | compartment. FASTKD2 is a mitochondrial matrix and mitochondrial nucleoid protein with TWO independent ECO:0000269 experimental citations and a sole HPA mitochondrial IF call; PSIP1/LEDGF is a nucleoplasmic chromatin protein. NOT the interaction statistic, which is fourth lowest of the 22 at p = 0.113. |
| hgnc ID | HGNC:29160 |
| symbol | FASTKD2 |
| gene name | FASTK mitochondrial RNA regulator 2 |
| ensembl gene ID | ENSG00000118246 |
| uniprot | Q9NYY8 |
| length AA | 710 |
| chromosomal location | 2q33.3 |
| entrez ID | 22868 |
| identifier provenance | HGNC REST fetch/symbol/FASTKD2 and UniProtKB Q9NYY8 retrieved live 2026-09-23; Ensembl id taken from the HGNC record. The previous symbol KIAA0971 was word-boundary matched against the full corpus id set and node text and resolves to no existing entry. |
| subcellular location hpa | Mitochondria (main and only, reliability Supported) |
| subnuclear location | NONE. FASTKD2 has no nuclear annotation at any evidence level in either resource. |
| prey localisation | PSIP1/LEDGF: nucleoplasm, HPA main, reliability Enhanced |
| cell type of the observation | K562 chronic myeloid leukaemia - a leukaemia cell, but BCR::ABL1 and not KMT2A-rearranged |
| cell state of the observation | over-expressed BioID2 fusion construct, one isoform per protein, proliferating line |
| epigenetic role | None. FASTKD2 is a mitochondrial RNA-granule protein required for mitochondrial ribosome biogenesis; it binds 16S rRNA and has a single RAP RNA-binding domain. |
| disease association | FASTKD2 mutations cause a mitochondrial encephalomyopathy associated with reduced cytochrome c oxidase activity |
| shown for fusion | NO - nobody has looked in any fusion. The row is wild-type PSIP1 in K562. |
| precedent | this graph has declined four PSIP1 partners and one KMT2A partner on compartment grounds already - GDF5, IGHMBP2, PPP1R13B, TMEM214 and TMCO1 - all from a single crosslinking study (PMID 30021884). This is the same verdict reached for the first time on a BioID2 proximity row, a different method with the same failure mode. |
| what would reverse this | demonstration of an extra-mitochondrial FASTKD2 pool in a haematopoietic cell, by fractionation or by imaging, which no published experiment has looked for |
| filed this way because | a gap list matching own_entry against gene-* and protein-* ids cannot see an interface-*-declined-* node; this beat measured that failure at 2 of 22 partners in its own class, so the symbol is in the title and aliases where a grep will find it |
prev and alias symbols
[ "KIAA0971" ]
domains
[
{
"name": "RAP",
"start": 634,
"end": 691
}
]the row
{
"core_protein": "PSIP1",
"core_uniprot": "O75475",
"partner_experimental_role": "bait",
"core_experimental_role": "prey",
"method": "MI:1314 proximity-dependent biotin identification (BioID2)",
"expansion": "MI:1060 spoke",
"mi_score": 0.27,
"imex": "IM-30059",
"pmid": "39251607",
"host_cell": "K562 (BCR::ABL1 CML; no KMT2A rearrangement)",
"n_rows": 1,
"n_publications": 1
}adjudication
{
"p_pair_both_margins_held": 0.113,
"rank_among_the_22": 4,
"bait_preys_recovered": 237,
"panel_median_preys_per_bait": 326,
"bait_selectivity_rank_in_panel": "33 of 50 (1 = widest labeller)",
"prey_baits_recovering_it": 9,
"of_panel": 50,
"note": "THE STATISTIC DOES NOT SUPPORT THIS DECLINE and the entry says so.",
"source": "analyses/interactome-tail-size-and-abundance-adjudication/pair_null.json"
}subcellular location uniprot
[ "Mitochondrion matrix, mitochondrion nucleoid (ECO:0000269, PMID 20869947 and PMID 25683715)", "Mitochondrion matrix (ECO:0000305, PMID 18771761 - curator inference)" ]
existing weak signal recorded so it is not reread as corroboration
{
"what": "DepMap co-dependency FASTKD2-PSIP1",
"r": 0.125,
"n_shared_lines": 1208,
"variance_explained_pct": 1.6,
"source_entry": "ana-depmap-co-dependency-cannot-rank-the-interactome-tail-either-and-my-pass-criterion-was-a-count-without-a-baseline",
"reading": "that entry's own conclusion is that its pass criterion was a count without a baseline; r = 0.125 over 1,208 lines is distinguishable from zero and indistinguishable from nothing"
}bounded absence
{
"queries": [
"FASTKD2[tiab] AND KMT2A[tiab] -> 0",
"FASTKD2[tiab] AND MLL[tiab] -> 0",
"FASTKD2[tiab] AND (PSIP1[tiab] OR LEDGF[tiab]) -> 0"
],
"querytranslations": [
"\"FASTKD2\"[Title/Abstract] AND \"KMT2A\"[Title/Abstract]",
"\"FASTKD2\"[Title/Abstract] AND \"MLL\"[Title/Abstract]"
],
"name_list_tried": [
"FASTKD2",
"PSIP1",
"LEDGF"
],
"name_list_NOT_tried": [
"KIAA0971"
],
"total_literature": "43 records for FASTKD2 in any title or abstract",
"bound": "titles and abstracts only; supporting figures not searched"
}- Declining a curated pair from the only modern proximity map in a leukaemia cell has a cost: if the decline is wrong, the graph loses one of very few rows measured in a haematopoietic context. The compartment evidence is strong enough to bear that cost and the entry states what would reverse it.
- FASTKD2 loss causes a human mitochondrial encephalomyopathy. If any future rationale reached for it as a target, the germline phenotype is the ceiling on that idea and should be read first.
- Does FASTKD2 have an extra-mitochondrial pool in a haematopoietic cell? No published experiment has looked, and a fractionation would either reverse this decline or close it permanently.
- How many of the 50 RBP baits in IM-30059 are mitochondrial, and do they systematically recover nuclear preys? If they do, that is a quantifiable method artefact in the deposition this project leans on most heavily, and it would bear on every row in it. FASTKD2 and TBRG4 are both FASTK-family mitochondrial proteins in the panel, which makes this a two-bait test rather than an anecdote.
- Two ribosome-biogenesis baits in this deposition - FASTKD2 (mitochondrial) and DDX55 (nucleolar) - both reach core proteins of this project and both scored NOT RESOLVED. Is ribosome-biogenesis machinery over-represented among the baits reaching KMT2A and PSIP1 relative to its share of the panel? This graph's cycle-129 test asked a version of that question and found no enrichment; adding the mitochondrial arm was not part of it.
- E1established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper
IntAct serves one curated human FASTKD2-PSIP1 row: PSI-MI MI:1314 proximity-dependent biotin identification, MI:1060 spoke expansion, FASTKD2 experimental role MI:0496 bait and PSIP1 MI:0498 prey, MI-score 0.27, host taxid:9606(human-k_562), IMEx IM-30059, publication PMID 39251607; FASTKD2 is one of the 50 RBP baits Khoroshkin et al. fused to BioID2 in K562 cells, and the authors state that RBP over-expression, BioID2 fusion and the use of a single isoform per protein may bias the data.
Systematic identification of post-transcriptional regulatory modules. · Khoroshkin M, Buyan A, Dodel M, Navickas A, Yu J, Trejo F, et al. · Nat Commun 2024What this evidence cannot show
Abstract read; full text not read. Row-level fields retrieved live from IntAct PSICQUIC MITAB 2.7 on 2026-09-23 (19,403 rows fetched, endpoint-declared count 19,403). Rule 93: PSICQUIC is a floor relative to IntAct REST and a REST query on Q9NYY8/O75475 was not run. Spoke expansion means the binary pair was never measured.
Evidence that argues against this (1)
- K562 is a leukaemia line, which is more than most of this project's interactome rows can say, and a bona-fide-pair curation step was applied by the authors before deposition. Declining a curated pair from a leukaemia cell is a real cost and this entry does not pretend the row is worthless - only that it cannot be an interaction as drawn. source 2026
- E2moderate confidence 3 of 5: One strong study, or several smaller consistent ones.dataset
In IM-30059 FASTKD2 recovered 237 of 2,681 human preys, ranking 33rd of 50 baits against a panel median of 326, while PSIP1 is recovered by 9 of the 50 baits; under a null holding both margins FASTKD2 recovering PSIP1 has p = 0.113, which is NOT RESOLVED under the pre-registered thresholds. Separately, PSIP1's 9-of-50 recovery is 1.2 times a negative-binomial prediction from its length and abundance and sits at the 66th percentile of 229 preys matched on both covariates.
What this evidence cannot show
20,000 draws, seed 20260923. Bonferroni over the 62 pairs tested is 0.0008 and nothing survives it. THE STATISTIC DOES NOT SUPPORT THIS DECLINE - p = 0.113 is fourth lowest of the 22 partners scored. The decline is on compartment.
Evidence that argues against this (1)
- PSIP1 at 9 of 50 baits is at the 66th percentile of its size- and abundance-matched peers - unremarkable, which cuts both ways. It means PSIP1 is not a promiscuous prey whose recovery can be dismissed wholesale, so declining one of its nine baits requires the specific compartment argument rather than a general appeal to promiscuity. source 2026
- E3established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper
FAST is the founding member of the FASTK domain-containing family FASTKD1-5; the mitochondrial localisation of FASTKD2 was confirmed and all FASTKD family members are found in mitochondria; mutations in FASTKD2 have been linked to a mitochondrial encephalomyopathy associated with reduced cytochrome c oxidase activity.
Fast kinase domain-containing protein 3 is a mitochondrial protein essential for cellular respiration. · Simarro M, Gimenez-Cassina A, Kedersha N, Lazaro JB, Adelmant GO, Marto JA, Rhee K, Tisdale S, Danial N, Benarafa C, Orduña A, Anderson P · Biochem Biophys Res Commun 2010What this evidence cannot show
Abstract read; full text not read. The paper's subject is FASTKD3; the FASTKD2 localisation is a confirmation reported alongside. It is one of the two ECO:0000269 citations UniProt gives for FASTKD2's matrix and nucleoid location and is cited here for that.
Evidence that argues against this (1)
- Confirming that a protein is mitochondrial does not exclude a minor extra-mitochondrial pool, and no experiment in either localisation paper looked for one. The decline is that nothing supports a nuclear pool, not that one is impossible. source 2026
- E4established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper
Mitochondrial RNA granules contain a large toolbox of proteins dedicated to RNA metabolism; of four uncharacterised putative RNA-binding proteins investigated, the helicases DHX30 and DDX28 and the FASTK-family protein FASTKD2 are required for mitochondrial ribosome biogenesis, and RNA-sequencing showed DDX28 and FASTKD2 bind the 16S rRNA. UniProt Q9NYY8 is 710 aa with a single annotated RAP domain at residues 634-691 and assigns mitochondrion matrix and mitochondrion nucleoid with ECO:0000269 evidence; the Human Protein Atlas assigns FASTKD2 a main subcellular location of Mitochondria, sole, at the Supported reliability tier.
Mitochondrial RNA Granules Are Centers for Posttranscriptional RNA Processing and Ribosome Biogenesis. · Antonicka H, Shoubridge EA · Cell Rep 2015What this evidence cannot show
Abstract read; full text not read. The UniProt and HPA records were fetched live on 2026-09-23. Work is in cultured human cells, not haematopoietic. The RAP domain is FASTKD2's only annotated domain and is an RNA-binding module, consistent with the mitochondrial RNA role and offering no nuclear function.
Evidence that argues against this (1)
- FASTKD2 is an rRNA-binding ribosome-biogenesis factor, and so is DDX55, another bait in the same deposition whose KMT2A row this beat also ranked NOT RESOLVED. If ribosome-biogenesis baits systematically pick up chromatin proteins in this map, the shared explanation is a method artefact rather than two separate biological stories - which argues for declining both rather than promoting either. source 2015
- E5established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database
PubMed returns zero records for FASTKD2 AND KMT2A in titles and abstracts, zero for FASTKD2 AND MLL, and zero for FASTKD2 AND (PSIP1 OR LEDGF), against a whole FASTKD2 literature of 43 records.
What this evidence cannot show
All zeros carry a non-empty querytranslation and a warninglist reading 'No items found.'. NAME LIST TRIED: FASTKD2, and for the prey both PSIP1 and LEDGF. NOT TRIED: KIAA0971, the HGNC previous symbol, which is the spelling in pre-2003 sequencing literature. Bounded by titles and abstracts only.
Evidence that argues against this (1)
- A 43-paper literature means the absence records that nobody has looked rather than that the relationship was sought and not found. The decline rests on the compartment evidence, which is positive evidence for a different location, not on this absence. source 2026
declined on compartment: mitochondrial matrix and nucleoid bait, nucleoplasmic prey, one BioID2 spoke row in K562
the same compartment-mismatch verdict, reached for the first time on a BioID2 proximity row rather than a crosslink
FASTKD2 is one of the 50 BioID2 baits in IM-30059; this node records its adjudication
the two ribosome-biogenesis baits in IM-30059 that reach a core protein - one mitochondrial, one nucleolar, both NOT RESOLVED by the statistic
PSIP1 at 9 of 50 baits is 1.2x predicted and the 66th percentile of a 229-member length- and abundance-matched pool, so declining one of its nine needs a specific argument