PSIP1 makes two proteins and only one of them has the IBD: a p52 result is not a p75 result, and the knockout mouse removes both
also called LEDGF p52 · LEDGF/p52 versus LEDGF/p75 · PSIP1 isoforms
The PSIP1 gene makes two different proteins, a long one and a short one. Only the long one has the part that grips the leukaemia protein. A lot of published LEDGF biology is about the short one and does not apply.
The same thing, for a scientist
PSIP1 encodes two coactivators from the same locus, **p75** and **p52**, described together in the founding 1998 cDNA paper. They share the N-terminal PWWP domain, the AT-hooks and the nuclear localisation signal; **p52 lacks the C-terminal integrase-binding domain entirely** and carries a short unique C-terminal tail instead. Every KMT2A-relevant contact - IBM1 at KMT2A 124-133, IBM2 at 147-152, the menin-bridged ternary complex, the CK2 phospho-switch, the HIV integrase competition - is on the **IBD and therefore exists only in p75**. Several of the most-cited LEDGF functions are p52 functions: p52 is the isoform shown to bind methylated H3K36 and splicing factors and to regulate alternative splicing. This matters practically in three places. Searching: a PubMed hit on 'p52' is not a hit on the leukaemic protein. Reading: the conditional and germline Psip1 knockouts remove **both** isoforms, so their phenotypes cannot be attributed to p75. And interpreting: an antibody or a knockdown construct targeting shared exons hits both.
What each isoform has
| Feature | p75 | p52 |
|---|---|---|
| PWWP (H3K36me2/3 reader) | yes | yes |
| AT-hooks, NLS, charged regions | yes | yes |
| IBD (integrase-binding domain, ~347-435) | yes | no |
| Binds KMT2A IBM1/IBM2 | yes | no |
| Binds menin-KMT2A ternary complex | yes | no |
| Binds HIV-1 integrase, POGZ, JPO2 | yes | no |
| Binds ASF/SF2 (SRSF1) and regulates alternative splicing | see below | yes, the characterised isoform |
| Unique C-terminal tail | no | yes |
The split is not tidy on splicing. p52 is the isoform characterised as binding methylated H3K36 together with splicing factors and contributing to alternative splicing regulation 2. But p75 also associates with splicing factors and contributes to exon choice in half the transcription units with alternative isoforms, and the same study used a cell line lacking the p75 IBD specifically to show it 3. So both isoforms touch splicing, by different routes, and an entry that attributes 'the splicing function' to one of them without saying which is wrong either way.
Three places this changes a conclusion
1. Literature counts. p52 as a search alias returns papers about a different protein entirely - NF-kB2 p52 is a major confound - as well as LEDGF/p52 papers that do not bear on the IBD. It was excluded from this beat's counted-absence queries for that reason, and the exclusion is recorded in absence-ledgf-removal-from-established-kmt2a-leukaemia.
2. Knockout interpretation. The Psip1 knockout mouse literature is explicit: knockout of Psip1 removes 'LEDGF/p75 and the shorter LEDGF/p52 isoform' 4. So perinatal lethality with homeotic skeletal transformation, and the haematopoietic phenotype of the conditional knockout, are Psip1-locus phenotypes, not p75 phenotypes. A p75-selective drug is not predicted by them. The one place isoform attribution has been done properly is the DNA-damage-response knockout series, where re-expression of p75 specifically rescued 5.
3. Chemoresistance. The AML chemoresistance signal was reported for 'the p75 and p52 splice variants' together, and in transfection both p75 and p52b antagonised daunorubicin-induced apoptosis while p52 variants lacking exon 6 were pro-apoptotic 6. An agent that lowers total PSIP1 output shifts that balance in an unpredictable direction.
The naming trap, stated for the checker
Aliases in circulation for the p75 protein: PSIP1, PSIP2, LEDGF, LEDGF/p75, DFS70, p75. Of these, p75 collides with p75NTR/NGFR and DFS70 is the autoantigen name used almost exclusively in the rheumatology literature. Aliases that are not the same protein: p52 (different isoform, no IBD) and NF-kB2 p52 (different gene). Any counted claim on this axis should say which alias set it used and which it excluded, because the recall and precision failures run in opposite directions.
| gene | PSIP1 |
|---|---|
| knockout removes | both isoforms |
isoforms
[
{
"name": "p75",
"has_IBD": true,
"kmt2a_relevant": true
},
{
"name": "p52",
"has_IBD": false,
"kmt2a_relevant": false
}
]shared domains
[ "PWWP", "AT-hooks", "NLS" ]
aliases same protein
[ "PSIP1", "PSIP2", "LEDGF", "LEDGF/p75", "DFS70", "p75" ]
aliases not same protein
[ "p52 (different isoform, no IBD)", "NF-kB2 p52 (different gene)", "p75NTR/NGFR (different gene)" ]
- A PWWP-directed agent cannot be isoform-selective, so it removes p52's splicing function alongside whatever it does to p75.
- p52 variants with and without exon 6 have opposite effects on apoptosis, so an agent that perturbs total PSIP1 output could shift the balance toward the pro-survival species.
- Every mouse phenotype cited as evidence that LEDGF is dispensable is a both-isoform phenotype, so the safety argument and the efficacy argument rest on different molecular entities.
- Is p52 expressed in KMT2A-rearranged blasts, and at what ratio to p75? If p52 is abundant it competes for H3K36me2/3 nucleosomes through the same PWWP domain, which would matter for any PWWP-directed agent and has never been measured in this disease.
- Would a p75-selective agent be possible at all? The only isoform-unique surfaces are the IBD (p75) and the short C-terminal tail (p52), so an IBD ligand is automatically p75-selective and a PWWP ligand is automatically not.
- Do the existing PSIP1 shRNA reagents used in the leukaemia literature target shared exons? If so, every published knockdown result on this axis is a dual-isoform result.
- E1established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper
PSIP1 encodes two transcriptional coactivators, p52 and p75, identified together as cDNAs from the same locus and reported as revealing an alternate regulatory mechanism of transcriptional activation.
Isolation of cDNAs encoding novel transcription coactivators p52 and p75 reveals an alternate regulatory mechanism of transcriptional activation. · Ge H, Si Y, et al. · EMBO J 1998What this evidence cannot show
The founding cDNA paper; cited for the existence and relationship of the two isoforms. Domain boundaries as used in this graph come from UniProt O75475 and the structural entries, not from this paper.
Evidence that argues against this (1)
- A 1998 cDNA description predates the domain nomenclature this graph uses, so the mapping of 'p52 lacks the IBD' onto the modern residue ranges is an inference from later structural work rather than a statement in this paper. Tesina P et al. 2015
- E2established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper
Psip1/Ledgf p52 binds methylated histone H3K36 and splicing factors and contributes to the regulation of alternative splicing - a function characterised on the p52 isoform.
Psip1/Ledgf p52 binds methylated histone H3K36 and splicing factors and contributes to the regulation of alternative splicing. · Pradeepa MM, Sutherland HG, et al. · PLoS Genet 2012What this evidence cannot show
The splicing characterisation is isoform-specific to p52 in this paper; it does not exclude a p75 contribution, which was later shown separately.
Evidence that argues against this (1)
- p75 also associates with splicing factors and contributes to exon choice, so treating splicing as a p52-only function is itself an error. Singh PK et al. 2015
- E3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
LEDGF/p75 associates with splicing factors and contributes to splicing patterns in half of the transcription units that have alternative isoforms, shown using cells lacking LEDGF/p75 or lacking the p75 integrase-binding domain specifically.
LEDGF/p75 interacts with mRNA splicing factors and targets HIV-1 integration to highly spliced genes. · Singh PK, Plumb MR, et al. · Genes Dev 2015What this evidence cannot show
HEK293T cells, in the context of an HIV integration-site study. The IBD-deletion line is the isoform-resolving reagent and is the reason this claim can be attributed to p75.
Evidence that argues against this (1)
- The splicing changes were measured in a cell line engineered for HIV integration studies, and the same paper's principal finding is about integration site selection, so the splicing analysis is secondary and not independently replicated. Pradeepa MM et al. 2012
- E4established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper
Knockout of the Psip1 gene removes both LEDGF/p75 and the shorter LEDGF/p52 isoform, so Psip1 knockout phenotypes cannot be attributed to p75 alone.
Embryonic Lethality Due to Arrested Cardiac Development in Psip1/Hdgfrp2 Double-Deficient Mice. · Wang H, Shun MC, et al. · PLoS One 2015What this evidence cannot show
Stated by the authors of the double-knockout study in describing the existing Psip1 knockout. Open access in Europe PMC (PMC4569352).
Evidence that argues against this (1)
- The original gene-trap allele produces a Psip1-betageo fusion retaining the PWWP and AT-hook domains while missing the conserved C terminus, so it is a C-terminal truncation rather than a null, and its phenotype is not simply 'both isoforms absent' either. Sutherland HG et al. 2006
- E5strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
Isoform attribution has been done correctly in at least one system: in a CRISPR LEDGF knockout with separate p75 and p52 re-expression lines, re-expression of LEDGF/p75 rescued the DNA-damage-response phenotypes.
LEDGF/p75 Is Required for an Efficient DNA Damage Response. · Liedtke V, Schröder C, et al. · Int J Mol Sci 2021What this evidence cannot show
One engineered cell background. Cited as the methodological example rather than for the repair biology itself.
Evidence that argues against this (1)
- Rescue by overexpressed p75 does not exclude a p52 contribution at endogenous levels, since an overexpressed isoform can cover a function it does not normally perform. Liedtke V et al. 2021
- E6moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper
In chemotherapy-resistant AML blasts, both the p75 and p52 splice variants were upregulated, and in transfection both LEDGF/p75 and p52b antagonised daunorubicin-induced apoptosis while p52 splice variants lacking exon 6 had pro-apoptotic effects.
LEDGF/p75 has increased expression in blasts from chemotherapy-resistant human acute myelogenic leukemia patients and protects leukemia cells from apoptosis in vitro. · Huang TS, Myklebust LM, et al. · Mol Cancer 2007What this evidence cannot show
Differential-display screening on pooled blasts with small numbers, plus transfection in an AML line and HEK-293. The opposing effects of different p52 variants make any total-PSIP1 manipulation hard to predict. Open access (PMC1876472).
Evidence that argues against this (1)
- Full-length LEDGF/p75 protected against truncated pro-apoptotic LEDGF/p75 in the same experiments, so the isoform balance is a system with internal opposition rather than a simple dose relationship. Huang TS et al. 2007
isoform disambiguation for this gene
the alias exclusion the counted queries rest on
the IBD exists only in p75
the allele removes both isoforms, so the safety argument is not isoform-resolved
opposing isoform effects on apoptosis
both isoforms touch splicing by different routes
the only subnuclear imaging dataset uses a pan-isoform antibody and cannot separate p75 from p52