Oncofusion

start / graph / target-cd15-lewis-x

Targetstrong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.Q48-surface-antigens

CD15 (Lewis-x, SSEA-1, 3-fucosyl-N-acetyllactosamine) - a sugar, not a protein: the most-named antigen in this graph that no gene encodes

also called CD15 · Lewis x · Lewis-x · Le-x · LeX · SSEA-1 · 3-FAL · 3-fucosyl-N-acetyllactosamine · Gal-beta-1,4-(Fuc-alpha-1,3)-GlcNAc · X-hapten

CD15 is not a protein at all - it is a small branched sugar stuck onto the outside of proteins and fats by enzymes, so no single gene makes it. It appears in 24 entries of this knowledge graph and could never have had an entry written by our automatic tool, because that tool looks up gene names.

The same thing, for a scientist

CD15 is the carbohydrate epitope **Gal-beta-1,4-(Fuc-alpha-1,3)-GlcNAc** - Lewis-x, also called SSEA-1 - displayed on both glycoproteins and glycosphingolipids. **No gene encodes it.** It is built post-translationally by alpha-1,3-fucosyltransferases acting on a type-2 N-acetyllactosamine precursor; in myeloid cells the relevant enzymes are **FUT4** (P22083, 530 aa) and **FUT9** (Q9Y231, 359 aa), and FUT9 is the enzyme whose crystal structures were solved generating Lewis-x and Lewis-y [E4][E5]. **Why it matters on this beat.** In adult B-precursor ALL, NG2/CSPG4 expression - which predicts *KMT2A* rearrangement with sensitivity 0.89, specificity 0.89 and PPV 0.93 - is significantly associated with a **CD10(-)/CD34(-)/CD24(-)/CD65s(+)/CD15(+)/CD13(-)/CD33(-)** phenotype, and **NG2 predicted KMT2A rearrangement with 100% accuracy among patients with a CD65s-positive and/or CD15-positive immunophenotype** [E1]. CD15 is therefore not a bystander marker: it is part of the phenotype that gives NG2 its discriminating power. **And it is the worked example of this graph's blind spot.** Measured 2026-09-20, CD15 appears in **24** node files, more than any other antigen without an entry, while FUT4 appears in 2 and FUT9 in 1. The named-but-unwritten detector enumerates HGNC protein-coding symbols, so a 24-entry topic was structurally invisible to it [E6]. **The trap inside the fix.** `CD15` **is** resolvable to a gene symbol - it is a **previous official HGNC symbol of FUT4** [E3]. So a naive repair that adds previous symbols to the queue does not fix the blind spot; it produces a confident **wrong** mapping, because CD15 the epitope is not the product of FUT4 and is made by several enzymes on many carriers. **Blindness and false resolution are both failure modes and the second is worse.**

Go deeper
What builds it, and why that is the bridge back to a gene-keyed instrument

Why this entry is type: "target" and its id does not begin gene-. The legal node types in data/graph/SCHEMA.md are fusion, gene, protein, domain, breakpoint, partner_interface, mechanism, cell_type, cell_state, disease, patient_story, epidemiology, drug, target, trial, resistance, pathway, noncoding_element, hypothesis, experiment, analysis, analogy, collaborator, concept and risk. There is no antigen type and no glycan type. target is the closest legal fit for something that is displayed on a cell surface and can be bound by an antibody. It is emphatically not a gene and an id of the form gene-cd15 would encode exactly the error this beat exists to fix - see concept-a-cd-number-is-not-a-gene.

Biosynthesis. An alpha-1,3-fucosyltransferase transfers fucose onto the GlcNAc of a type-2 LacNAc acceptor. FUT9 is a dedicated Lewis-x/Lewis-y synthase whose structures with GDP, acceptor glycans and as a donor-analogue Michaelis complex were determined in 2023 4; FUT4 is the myeloid-specific enzyme, HGNC name fucosyltransferase 4, alias name alpha (1,3) fucosyltransferase, myeloid-specific 3. Independent regulation of Fuc-TIV (FUT4) and Fuc-TVII (FUT7) modulates cell-surface antigen expression as myeloid cells develop 7.

The one real bridge to KMT2A biology found on this beat. A co-expression network analysis of 204 pre-treated paediatric ALL patients identified a module positively correlated with MLL rearrangement, whose hub genes were PROM1, LGALS1, CD44, FUT4 and HOXA10, all significantly increased in MLL-rearranged samples and validated independently 2. That is FUT4 - the CD15-building enzyme - as a hub of the KMT2A-rearranged expression programme. It is the only result located on this beat that connects the CD15 epitope to a gene a dependency screen, a DepMap query or a CRISPR experiment can reach.

The retrieval asymmetry, measured. On PubMed, 2026-09-20: CD15 AND (KMT2A OR MLL) returns 36 records; (FUT4 OR FUT9) AND (KMT2A OR MLL) returns 1 - PMID 32347296, the co-expression paper above. Controls: FUT4 OR FUT9 alone returns 350, KMT2A OR MLL alone 9,229. The epitope is discussed thirty-six times more often than the enzymes that make it 6.

What CD15 positivity does and does not tell you
  • It is a maturation marker. The SSEA-1/CD15 epitope is restricted to normal and leukaemic myeloid cells, carried by several neutral glycosphingolipids with seven to more than ten monosaccharides, and expressed by myeloid cells at all maturity levels with quantitative variation; myeloblasts stain more weakly than promyelocytes, myelocytes and metamyelocytes 8. In AML, CD15 was rarely detected in the CD65s-low, minimally differentiated M0/M1 group 9.
  • On a B-lineage blast it is aberrant, and that is the signal. CD15 on a CD19-positive blast is lineage infidelity, and in combination with CD65s it is what makes NG2 a perfect predictor in 1.
  • It is not a therapeutic target here. Nothing read on this beat proposes a CD15-directed therapy in KMT2A-rearranged leukaemia, and a carbohydrate shared with normal neutrophils would be a poor one.
  • A flow result is antibody-dependent in a way a protein antigen's is not. Different anti-CD15 clones can recognise the same terminal epitope on different carriers, and the CD65/CD65s pair is the documented case where clones sharing a CD designation turned out to have two different antigen specificities [see target-cd65-vim2-neutral-glycolipid].
Facts and figures
antigen classcarbohydrate epitope (terminal trisaccharide), displayed on both glycoproteins and glycosphingolipids
chemical structureGal-beta-1,4-(Fuc-alpha-1,3)-GlcNAc (Lewis-x / SSEA-1 / 3-fucosyl-N-acetyllactosamine)
encoded by a genefalse
why not a genepost-translational glycan built by alpha-1,3-fucosyltransferases on a type-2 LacNAc acceptor; several enzymes can produce it and it is carried on many different molecules
kmt2a relevancecomponent of the CD10(-)/CD34(-)/CD24(-)/CD65s(+)/CD15(+)/CD13(-)/CD33(-) phenotype in which NG2 predicts MLL rearrangement with 100% accuracy in adult B-precursor ALL
graph mentions 2026 09 20
FUT91.000FUT9: 1FUT42.000FUT4: 2CD1524.0CD15: 24
synthesising enzymes
[
  {
    "gene": "FUT4",
    "uniprot": "P22083",
    "length_aa": 530,
    "note": "myeloid-specific alpha(1,3)fucosyltransferase; CD15 is a PREVIOUS HGNC symbol of this gene"
  },
  {
    "gene": "FUT9",
    "uniprot": "Q9Y231",
    "length_aa": 359,
    "note": "dedicated Lewis-x/Lewis-y synthase; crystal structures solved 2023"
  }
]
graph mentions 2026 09 20 — the numbers behind the chart
{
  "CD15": 24,
  "FUT4": 2,
  "FUT9": 1
}
pubmed 2026 09 20
{
  "CD15 AND (KMT2A OR MLL)": 36,
  "(FUT4 OR FUT9) AND (KMT2A OR MLL)": 1,
  "controls": {
    "FUT4 OR FUT9": 350,
    "KMT2A OR MLL": 9229
  }
}
How this could bite us back
  • **Do not build a CD15-directed intervention from this entry.** CD15 is abundant on normal neutrophils [E8], so a CD15-directed agent has the granulocyte problem of CLEC12A without any of its leukaemic-stem-cell selectivity.
  • **Targeting the enzyme is not targeting the epitope.** Even if FUT4 knockdown removed CD15 from blasts, FUT4 is the myeloid-specific fucosyltransferase and its loss would strip the epitope from normal myeloid cells too, with unknown consequences for selectin-mediated trafficking.
  • **The repair for the blind spot can be worse than the blind spot.** Adding HGNC previous symbols to the named-but-unwritten queue would map CD15 to FUT4 and produce a confident wrong gene assignment for a 24-entry antigen [E3].
Still open
  1. Which alpha-1,3-fucosyltransferase actually builds the CD15 seen on KMT2A-rearranged blasts? FUT4 is a hub gene of the MLL-r co-expression module [E2] but nobody has knocked it out and asked whether surface CD15 disappears. That single experiment would convert a 24-entry antigen from unreachable to dependency-screenable.
  2. Does CD15 sit on a glycoprotein or a glycosphingolipid on KMT2A-rearranged B-lineage blasts? Macher 1990 found the epitope on neutral glycosphingolipids in myeloid cells [E8]; the carrier determines whether any protein-level approach can reach it.
  3. Is the CD15/CD65s pair on a KMT2A-rearranged blast a fusion-driven programme or a passive consequence of the blast's maturation arrest? The two predict opposite things about whether the antigens persist under menin inhibition.
Every source behind this entry (9)
  1. E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    In 313 adult ALL patients with MLL rearrangement determined by RT-PCR and FISH in 120 B-precursor cases, NG2 expression was significantly associated with a CD10(-)/CD34(-)/CD24(-)/CD65s(+)/CD15(+)/CD13(-)/CD33(-) phenotype and showed sensitivity 0.89, specificity 0.89 and positive predictive value 0.93 for MLL rearrangement; NG2 predicted MLL rearrangement with 100% accuracy among patients with a CD65s-positive and/or CD15-positive immunophenotype.

    PMID 12886247DOI
    What this evidence cannot show

    Abstract read in full from raw NCBI efetch XML on 2026-09-20; full text not read. **Adults, not children and not infants** - 313 adult ALL patients. The '100% accuracy' figure is conditioned on a CD65s+/CD15+ subgroup whose size is not given in the abstract, and a perfect accuracy in an unstated denominator is exactly the statistic that fails to replicate. The full phenotype includes **CD34(-), CD13(-) and CD33(-)**, which are routinely dropped when this result is quoted as 'CD10-/CD24-/CD65s+/CD15+'; this entry states the full string. The same paper reports NG2 positive in 3 patients without detectable MLL rearrangement and negative in 8 patients with MLL-AF4 transcripts.

    Evidence that argues against this (1)
    • The paper's own discordances are the counter-evidence: 8 MLL-AF4-positive patients were NG2-negative. So the marker combination that reaches 100% accuracy in one direction has a real false-negative rate in the other, and CD15/CD65s positivity is doing work only within the NG2-positive population. Xiu Y et al. 2018
  2. E2moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

    In a gene co-expression network built from 204 pre-treated paediatric ALL patients, the module most positively correlated with MLL rearrangement had PROM1, LGALS1, CD44, FUT4 and HOXA10 as hub genes, whose expression levels were significantly increased in MLL-rearranged samples and were validated in an independent investigation.

    PMID 32347296DOI
    What this evidence cannot show

    Abstract read in full from raw NCBI efetch XML on 2026-09-20; full text not read. **This is a co-expression / WGCNA analysis of public expression data, not a functional experiment**: hub-gene membership is a network-topology statement, not evidence that FUT4 does anything in KMT2A-rearranged leukaemia. The abstract does not state the MLL-rearranged fraction of the 204 patients, the partner-gene composition, or the platform. 'Validated by an independent investigation' is not described in the abstract.

    Evidence that argues against this (1)
    • FUT4 mRNA being high in MLL-rearranged samples does not establish that FUT4 is the enzyme producing the CD15 seen on those blasts - FUT9 and other alpha-1,3-fucosyltransferases can make the same epitope, and transcript level does not predict glycan display, which depends on acceptor availability and competing sialylation. The bridge from FUT4 to CD15 in this disease is unproven. Kadirvelraj R et al. 2023
  3. E3established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database

    HGNC record HGNC:4015 gives FUT4 as the approved symbol for 'fucosyltransferase 4', alias name 'alpha (1,3) fucosyltransferase, myeloid-specific', UniProt P22083, and lists CD15 among its PREVIOUS official symbols.

    HGNC record HGNC:4015 (FUT4) · HUGO Gene Nomenclature Committee · HGNC 2026
    What this evidence cannot show

    Fetched live from the HGNC REST API on 2026-09-20 (`/fetch/symbol/FUT4`). A previous symbol is a nomenclature fact about an enzyme gene; it is **not** a statement that the gene encodes the CD15 epitope, and the biology says it cannot, because CD15 is a post-translational carbohydrate made by several enzymes on many carriers. Recording this here is a warning, not an equivalence.

    Evidence that argues against this (1)
    • An HGNC free-text search on the string 'CD15' returns FUT4 with a high score, so any symbol resolver that consults previous symbols or free-text search will silently map the CD15 epitope onto one fucosyltransferase gene. That is a false resolution, and it is a more dangerous defect than the blindness it appears to cure. HUGO Gene Nomenclature Committee 2026
  4. E4established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper

    Crystal structures of human FUT9, an alpha-1,3-fucosyltransferase that generates Lewis-x and Lewis-y antigens, were determined in complex with GDP, with acceptor glycans, and as a FUT9-donor-analogue-acceptor Michaelis complex, revealing substrate specificity determinants supported by kinetic analysis of active-site mutants.

    Structural basis for Lewis antigen synthesis by the alpha1,3-fucosyltransferase FUT9. · Kadirvelraj R, Boruah BM, Wang S, Chapla D, Huang C, Ramiah A, et al. · Nat Chem Biol 2023
    PMID 37202521DOI
    What this evidence cannot show

    Abstract read in full from raw NCBI efetch XML on 2026-09-20; full text not read. Structural enzymology on recombinant protein; nothing in this paper concerns leukaemia, myeloid cells or KMT2A. It is cited for the biochemical fact that Lewis-x is an enzymatic product, which is the premise of this whole entry.

    Evidence that argues against this (1)
    • FUT9 is one of several GT10-family alpha-1,3/4-fucosyltransferases, and the paper's own framing is that Lewis antigens are synthesised by a COLLECTION of them with modular donor- and acceptor-binding specificity. Attributing myeloid CD15 to any single enzyme is therefore not supported by structure alone. Xiu Y et al. 2018
  5. E5established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database

    The reviewed human UniProt records are FUT4 = P22083 (530 aa, primary gene name FUT4, synonyms ELFT and FCT3A) and FUT9 = Q9Y231 (359 aa, primary gene name FUT9, no listed synonyms), both annotated as 4-galactosyl-N-acetylglucosaminide 3-alpha-L-fucosyltransferase (EC 2.4.1.152).

    UniProtKB reviewed entry P22083 (FUT4_HUMAN) · UniProt Consortium · UniProtKB 2026
    What this evidence cannot show

    Fetched live from the UniProt REST API on 2026-09-20 with fields accession, gene_primary, gene_synonym, protein_name, length. Each `gene_exact` query returned exactly one reviewed human row whose primary gene name equalled the queried symbol, per rule 62. The two proteins differ by 171 residues (530 vs 359), so unusually for this beat the length check is a real discriminator here.

    Evidence that argues against this (1)
    • A UniProt length is a canonical-isoform fact. For a Golgi type-II membrane glycosyltransferase the catalytic domain is a fraction of the chain and soluble secreted forms are produced by proteolysis, so the full-length number describes the record and not necessarily the active species in a cell. UniProt Consortium 2026
  6. E6strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.dataset

    As of 2026-09-20, CD15 appears in 24 of this graph's 1,465 node files while FUT4 appears in 2 and FUT9 in 1; and on PubMed `CD15 AND (KMT2A OR MLL)` returns 36 records while `(FUT4 OR FUT9) AND (KMT2A OR MLL)` returns 1.

    Q48 surface-antigen beat: bounded counted-absence sweep, PubMed E-utilities and Europe PMC REST · Q48-surface-antigens · Oncofusion knowledge graph (internal measurement, 2026-09-20) 2026
    What this evidence cannot show

    In-graph counts measured 2026-09-20 by Python regex over all node JSON files (patterns `\bCD15\b`, `\bFUT4\b`, `\bFUT9\b`); the `\b` boundary prevents CD15 matching CD150 but the count includes citation titles, so it slightly over-counts substantive use. The dispatch brief quoted 23 for CD15; this independent run gives 24, and the one-entry difference is not reconciled - **the correct reading is 'about two dozen', not either exact number**. PubMed counts from E-utilities esearch on 2026-09-20 with `querytranslation` read back; **rule 67 controls passed**: `FUT4 OR FUT9` alone = 350, `KMT2A OR MLL` alone = 9,229, so the instrument returns ones. Note that PubMed translates the bare token `CD15` to include `"lewis x antigen"[Supplementary Concept]`, which is why the 36 is a generous rather than a conservative count.

    Evidence that argues against this (1)
    • The 36-vs-1 ratio is partly an artefact of what each literature is ABOUT: flow-cytometry papers name antigens and glycobiology papers name enzymes, and the two fields rarely overlap regardless of any instrument's blindness. The ratio measures a field boundary as much as it measures a retrieval gap. Q48-surface-antigens 2026
  7. E7moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

    Across four leukaemic cell lines representing successive myeloid maturation stages (KG1a, KG1, HL-60 and induced HL-60), Fuc-TIV (FUT4) mRNA increased progressively to peak at the HL-60 promyelocyte stage, and that peak correlated with increased fucosyltransferase activity towards non-sialylated acceptors, transient downregulation of surface sialyl-Lewis-x, and upregulation of Lewis-x (CD15), VIM-2 (CD65s) and Lewis-y - so the surface levels of these antigens track independent regulation of the Fuc-TIV and Fuc-TVII genes.

    PMID 9376686DOI
    What this evidence cannot show

    Abstract read in full from raw NCBI efetch XML on 2026-09-20; full text not read. **Cell lines only** - KG1a, KG1 and HL-60 are leukaemic lines taken as proxies for maturation stages, and none is KMT2A-rearranged in any way this abstract states. mRNA and surface antigen are correlated, not causally linked by knockdown. **A DOI written into this block from inference before the record was fetched was wrong (10.1093/glycob/7.7.973 against the true 10.1093/glycob/7.6.835); the value here was copied from the fetched efetch record.** That near-miss is logged in this beat's memo as a live instance of the brief's identifier rule.

    Evidence that argues against this (1)
    • This is the clearest published link between FUT4 and both CD15 and CD65s, and it also undercuts any simple 'FUT4 makes CD15' story: the same FUT4 peak simultaneously RAISES Lewis-x and VIM-2 while LOWERING sialyl-Lewis-x, because the enzymes compete for shared acceptors. Glycan surface levels are a flux outcome, not a gene-dosage readout. Kadirvelraj R et al. 2023
  8. E8strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    Anti-SSEA-1, which binds glycoconjugates carrying the Gal-beta-1,4-(Fuc-alpha-1,3)-GlcNAc epitope, stains normal and leukaemic myeloid cells only; among normal marrow myeloid precursors myeloblasts stain more weakly than promyelocytes, myelocytes and metamyelocytes; the SSEA-1 epitope is carried by several neutral glycosphingolipids with seven to more than ten monosaccharides expressed by all myeloid cells regardless of maturity; and lymphocytic leukaemia marrows were negative.

    PMID 1690317DOI
    What this evidence cannot show

    Abstract read in full from raw NCBI efetch XML on 2026-09-20; full text not read. 1990 immunohistochemistry and thin-layer-chromatography immunostaining, not flow cytometry, and sample numbers are not given in the abstract. 'Lymphocytic leukaemia marrows were negative' predates the recognition of KMT2A-rearranged pro-B ALL as a myeloid-antigen-co-expressing entity and should not be read as contradicting [E1] - the cases in [E1] are exactly the exception.

    Evidence that argues against this (1)
    • This paper's negative in lymphocytic leukaemia is in direct tension with CD15 positivity being a defining feature of KMT2A-rearranged B-lineage ALL [E1]. The resolution is almost certainly that KMT2A-rearranged pro-B cases are rare and were not separately identified in 1990, but the tension is real and is recorded rather than smoothed over. Schwartz S et al. 2003
  9. E9established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper

    In 711 AML patients from seven ECOG trials, the 198 (28%) with low CD65s expression were enriched in minimally differentiated FAB M0/M1 morphology and in the early precursor antigens CD34, CD117 and terminal transferase, while the more mature myeloid antigens CD15 and CD11b were rarely detected in that group (all P < 0.0001).

    PMID 12886241DOI
    What this evidence cannot show

    Abstract read in full from raw NCBI efetch XML on 2026-09-20; full text not read. AML in adults (1986-1999 trials), not paediatric and not KMT2A-stratified - the abstract reports no difference between CD65s-low and CD65s-high groups in the distribution of cytogenetic prognostic groups, so nothing here is specific to KMT2A rearrangement. Cited for the maturation-stage relationship between CD15, CD65s and TdT, which is the reason all three appear together in this beat.

    Evidence that argues against this (1)
    • The inverse relationship between CD15/CD65s and TdT in AML is the opposite pairing to the KMT2A-rearranged B-ALL phenotype, where CD15+/CD65s+ blasts are frequently TdT-negative [see `gene-dntt`]. Maturation-antigen logic derived from AML does not transfer cleanly to KMT2A-rearranged B-lineage disease. Klairmont MM et al. 2021
Connected to