S1PR4 - the third parenchymally clean surface protein in the screen, a leukocyte-restricted GPCR, and not fusion-specific either
also called S1P4 · EDG6 · Edg-6 · sphingosine 1-phosphate receptor 4
S1PR4 is a receptor on the surface of white blood cells that helps them move around the body. Like the other two 'clean' proteins this screen found, it is absent from normal organs only because it belongs to white blood cells, so it is not specific to this leukaemia.
The same thing, for a scientist
S1PR4 (UniProt **O95977**, S1PR4_HUMAN, **384 aa**, ENSG00000125910) is a G-protein-coupled receptor for sphingosine 1-phosphate, restricted to the lymphoid and haematopoietic compartment [E1]. In the screen it is the third parenchymally clean gene: B-arm mean **140.45 nTPM** (SEM 44.8, RS4;11 236.1), parenchymal maximum **4.3 nTPM in adipose tissue**, with a PTPRC profile correlation of **0.811** [E2]. **It fails fusion-enrichment on both ratios**: 1.75x a B-precursor comparator (threshold 4x) and 2.45x the KMT2A-other-partner arm (threshold 2x, which it clears). GTEx puts it at **252.82 TPM in whole blood** with a whole-body median of **1.11 TPM** - the cleanest non-blood profile of the three, and again a restatement of leukocyte restriction [E3]. Normal spleen is 58.6 and bone marrow 18.5 nTPM [E2]. A GPCR is an unusual antibody target - seven transmembrane passes leave short extracellular loops - and this is a practical objection independent of specificity. **Recorded as a negative candidate.** PubMed returns **0** records for `S1PR4[tiab] AND leukemia[tiab]`, so nothing in the literature connects it to this disease [E4].
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.
I have never read one of these pictures before
A protein is one long chain of amino acids that folds up. The ribbon traces that chain from one end to the other; it is not a wire, it is a simplification of where the backbone runs. Drag to rotate, scroll to zoom.
Domains are the stretches of chain that do a particular job, and they are what the coloured blocks above the picture mark out. Click one and the view flies to it. In this project the important question about any domain is usually whether the fusion still has it — the labels say lost where it does not.
Anything in yellow sticks is a bound small molecule, which is usually the drug.
| uniprot | O95977 |
|---|---|
| length AA | 384 |
| ensembl | ENSG00000125910 |
| alphafold ID | AF-O95977-F1 |
| hpa protein class | G-protein coupled receptors, Predicted membrane proteins |
| b arm mean nTPM | 140.45 |
| b precursor comparator nTPM | 80.36 |
| ratio over b precursor | 1.75 |
| ratio over other kmt2a partner | 2.45 |
| parenchymal max nTPM | 4.3 |
| parenchymal peak tissue | adipose tissue |
| lymphoid profile r vs PTPRC | 0.811 |
| gtex v8 median of 54 tissues | 1.11 |
| screen verdict | F1 surface PASS, F2b expressed PASS, F4b parenchymal CLEAN, F3b fusion-enriched FAIL (1.75x B-precursor comparator, threshold 4x) |
| shown for fusion | NO - nobody has looked in any fusion. PubMed returns 0 for S1PR4[tiab] AND leukemia[tiab]; the legacy aliases EDG6 and S1P4 were not separately queried. |
index arm nTPM — the numbers behind the chart
{
"SEM": 44.8,
"RS4;11": 236.1,
"MV4-11": 0.1
}normal lymphoid nTPM — the numbers behind the chart
{
"spleen": 58.6,
"bone marrow": 18.5,
"lymph node": 15.5,
"appendix": 10.1,
"thymus": 4.5,
"tonsil": 4.5
}gtex v8 peak
{
"tissue": "Whole_Blood",
"tpm": 252.82
}- S1PR4 peaks in whole blood and sits at 18.5 nTPM in normal bone marrow, so its clean parenchymal score is an artefact of excluding exactly the compartment where it lives.
- S1P receptors govern lymphocyte trafficking, and the S1P axis is already drugged systemically. Perturbing S1PR4 would have immunological consequences well beyond depleting the cells that carry it.
- This entry's bounded absence omits two legacy aliases, which is the exact failure this project has recorded three cycles running. It is flagged rather than fixed because fixing it was out of scope for this beat, and it should not be quoted until it is.
- Do the legacy aliases EDG6 and S1P4 return leukaemia literature that the modern symbol misses? This beat did not query them and the bounded absence above is incomplete without it.
- Why is S1PR4 5-fold higher in RS4;11 than in SEM? Two KMT2A::AFF1 B-ALL lines disagreeing by that much on a supposedly lineage-restricted receptor is worth explaining before the mean is used for anything.
- Is there any precedent for a GPCR as a successful CAR target in haematological malignancy? GPRC5D in myeloma is the obvious comparator and was not examined here.
- E1established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database
S1PR4 is UniProt O95977 (S1PR4_HUMAN), sphingosine 1-phosphate receptor 4, 384 amino acids, a G-protein-coupled receptor annotated Cell membrane with extracellular topological domains.
What this evidence cannot show
UniProt release 2026_03; topological domains for GPCRs are inferred from the seven-transmembrane architecture rather than measured.
Evidence that argues against this (1)
- A seven-transmembrane receptor presents only short extracellular loops, which makes it a poor antibody and CAR target regardless of how specific its expression is. UniProt 2026
- E2moderate confidence 3 of 5: One strong study, or several smaller consistent ones.dataset
S1PR4 is 44.8 nTPM in SEM and 236.1 in RS4;11 (B-arm mean 140.45) with a parenchymal maximum of 4.3 nTPM in adipose tissue; it is 1.75x a B-precursor comparator and 2.45x the KMT2A-other-partner arm; PTPRC profile correlation 0.811; normal spleen 58.6 and bone marrow 18.5 nTPM.
analyses/kmt2a-aff1-surface-target-screen/ - PREREGISTRATION.md, run.py and results.json, this repository. Pre-registered before any expression value was inspected. · Oncofusion Knowledge Graph 2026What this evidence cannot show
Two index lines with a 5-fold spread between them (44.8 versus 236.1), and 0.1 nTPM in MV4-11. A B-arm mean over two such discordant values is a weak summary.
Evidence that argues against this (1)
- Bone marrow at 18.5 nTPM means S1PR4-directed therapy acts on the normal haematopoietic compartment, which for this antigen is not merely the accepted cost but the main site of expression. Human Protein Atlas 2026
- E3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.database
GTEx v8 puts S1PR4 at 252.82 TPM in whole blood with a whole-body median of 1.11 TPM across 54 tissues.
GTEx v8 median gene expression, S1PR4 (ENSG00000125910.5) across 54 tissues: Whole_Blood 252.82 TPM, whole-body median 1.11 TPM. Fetched 2026-09-23 · GTEx Portal 2026What this evidence cannot show
Whole blood is the peak tissue, which confirms leukocyte restriction rather than tumour specificity.
Evidence that argues against this (1)
- A whole-blood peak is the worst possible profile for an antigen intended to spare normal haematopoiesis, and it is invisible to a parenchymal maximum by construction. GTEx Portal 2026
- E4strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.database
PubMed returns 0 records for S1PR4[tiab] AND leukemia[tiab] on 2026-09-23.
PubMed esearch via scripts/pubmed_safe.py; S1PR4[tiab] AND leukemia[tiab] -> 0, querytranslation "S1PR4"[Title/Abstract] AND "leukemia"[Title/Abstract]; run 2026-09-23 · NCBI PubMed 2026What this evidence cannot show
A non-empty querytranslation was returned, so this is a genuine zero and not an instrument failure. Bounded by titles and abstracts; the legacy aliases EDG6 and S1P4 were NOT separately queried, which this project's own rules require and which is a gap in this bounded absence.
Evidence that argues against this (1)
- EDG6 and S1P4 are the older names and papers from before the S1PR nomenclature change would use them. This zero is therefore a floor on the modern symbol only and should not be quoted as 'nothing is known'. UniProt 2026
one of three parenchymally clean genes, and the one whose cleanliness is most clearly an exclusion artefact (whole blood 252.82 TPM)
- expressed in →KMT2A::AFF1 (MLL-AF4), t(4;11)(q21;q23)
44.8 nTPM in SEM and 236.1 in RS4;11, 1.75x a B-precursor comparator - not fusion-specific