LPP (lipoma-preferred partner) - a KMT2A fusion partner in acute monoblastic leukaemia whose day job is adhesion mechanotransduction, and which is barely expressed in leukocytes

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.

• THIS ENTRY NOMINATES NOTHING AND SUPPORTS NO INTERVENTION. LPP appears in one unreplicated 2001 case of therapy-related AML-M5 in an adult; it is not part of the infant KMT2A::AFF1 biology this graph exists to map, and nothing here identifies a target, a drug or a patient population.
• The specific misreading to guard against is importing LPP's solid-tumour mechanotransduction literature into leukaemia. LPP is a Src substrate in invadopodia and a tension sensor at focal adhesions in adherent cells - and it is barely expressed in leukocytes. 'Src substrate' plus 'leukaemia fusion partner' would suggest a kinase-inhibitor rationale; there is no evidence for one, the two literatures have no established point of contact, and treating the adhesion biology as relevant to a blast in suspension would be unfounded.
• The nomenclature resolution in E3 must not be reported as a KMT2A fusion retaining methyltransferase ACTIVITY. The retained CXXC/MT domain is a CpG-binding domain named for sequence homology to DNA methyltransferases; it is not a methyltransferase and transfers no methyl group. The catalytic SET domain is lost, as this graph's model holds.

1. Has any MLL-LPP protein ever been detected? Both 2001 papers stop at the transcript, the chimera's composition is predicted from sequence, and 25 years later there appears to be no antibody-level or mass-spectrometry evidence that the protein exists.
2. Does the intron-8 breakpoint matter? It is upstream of the intron 9-11 breakpoint cluster typical of KMT2A::AFF1, so MLL-LPP retains the same CXXC/MT domain but a shorter MLL contribution. Whether breakpoint position within the BCR changes the retained N-terminus's function has not been tested in this graph.
3. Why is a gene with 'no or very low expression in peripheral leukocytes' available as a fusion partner at all? A translocation requires both loci to be in proximity, and partner choice in KMT2A rearrangement is usually discussed in terms of nuclear architecture rather than expression - so a poorly expressed partner is either evidence against expression-driven partner selection or evidence that the UniProt tissue note is too coarse.
4. Does LPP appear in any KMT2A partner census not indexed by gene symbol? The 'two papers' figure is a lower bound from a tiab query, and pre-2000 cytogenetic literature indexes by chromosome band. A search over 3q28 and 11q23 band notation rather than gene tokens has not been run.
5. How many other apparent hits in this graph are index-record disagreements? The PMID 42345200 case shows a literal [tiab] query returning a record that contains the token in no retrieval route, and no guard here runs in that direction. The corpus has never been swept for cited PMIDs whose stated supporting token is absent from the fetched record.

1. E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.database

   UniProt Q93052 (LPP_HUMAN, 'Lipoma-preferred partner', 612 aa, no gene synonyms) records four chromosomal aberrations involving LPP, one of which is a KMT2A fusion: translocation t(3;11)(q28;q23) with KMT2A/MLL1 in acute monoblastic leukaemia. The other three are t(3;12) rearrangements with HMGA2 in lipomas, pulmonary chondroid hamartomas and parosteal lipomas. The protein carries three LIM zinc-binding domains at 414-473, 474-534 and 535-603. Its four recorded subcellular locations - Nucleus, Cytoplasm, Cell junction, Cell membrane - all carry NO evidence code, and UniProt adds that LPP 'Shuttles between the cytoplasm and the nucleus'. The tissue note states expression in a wide variety of tissues but 'no or very low expression in brain and peripheral leukocytes'.

   UniProtKB Q93052 (LPP_HUMAN), Lipoma-preferred partner - queried as gene_exact:LPP AND organism_id:9606 AND reviewed:true, which returned exactly one reviewed record
   ▸What this evidence cannot show

   The four subcellular locations are UNCODED, so UniProt itself attaches no experimental evidence to them, and the entry's FUNCTION text is hedged throughout ('May play a structural role', 'may also be implicated', 'Also suggested to serve as a scaffold protein') - this is a curator's synthesis of inference, not a measured localisation. All of it describes WILD-TYPE LPP; nothing here has been shown for the chimera. The single-reviewed-record assertion guards against picking the wrong accession but not against the record being incomplete.

   ▸Evidence that argues against this (1)

   • A UniProt disease note is a literature pointer, not independent weight. This particular note rests on the SAME single 2001 case report cited in E2 - so 'UniProt records the KMT2A fusion' and 'one patient has been described' are the same fact counted twice, and the entry should not be read as two sources agreeing.
2. E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

   A t(3;11)(q28;q23) was identified in a patient who developed AML-M5 three years after treatment for follicular lymphoma, and LPP on 3q28 was identified as the MLL fusion partner by inverse PCR. The breakpoint occurred in intron 8 of both MLL and LPP; two in-frame MLL-LPP transcripts fusing MLL exon 8 to LPP exon 9 were detected, the smaller carrying a 120 bp deletion of MLL sequence, and a reciprocal LPP-MLL transcript was also detected. The abstract states the predicted MLL-LPP fusion protein 'includes the A/T hook motifs and methyltransferase domain of MLL joined to the two last LIM domains of LPP'.

   Human LPP gene is fused to MLL in a secondary acute leukemia with a t(3;11) (q28;q23). · Dahéron L, Veinstein A, Brizard F, et al. · Genes, Chromosomes & Cancer 2001
   PMID 11433529DOI
   ▸What this evidence cannot show

   A SINGLE PATIENT, reported once in 2001 and never replicated - and a THERAPY-RELATED leukaemia arising after treatment for follicular lymphoma, so it is a secondary AML in an adult, not the de novo infant B-ALL biology this graph is built around. Everything is transcript-level: no MLL-LPP protein was detected, and the fusion protein's composition is 'predicted' from sequence. The relevance of this entry to KMT2A::AFF1 is comparative, not direct.

   ▸Evidence that argues against this (2)

   • LPP is essentially absent from the modern KMT2A fusion-partner literature. Large KMT2A-rearrangement partner surveys enumerate dozens of partners without LPP appearing as a recurrent one, and this project's own search finds only two primary papers, both from 2001 - so a single unreplicated case is compatible with a passenger rearrangement or a private event rather than a bona fide recurrent oncogenic fusion.
   • LPP having 'no or very low expression in peripheral leukocytes' (E1) cuts against the fusion mattering mechanistically. Most characterised KMT2A partners are nuclear proteins expressed in haematopoietic cells that contribute transcriptional-elongation function; a partner barely transcribed in the lineage, contributing two adhesion LIM domains, fits the model in which the partner's identity matters less than the loss of MLL's C-terminus - which would make LPP uninformative about partner-specific mechanism.
3. E3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

   The apparent contradiction between the Daheron abstract's 'methyltransferase domain of MLL' and this graph's model that every KMT2A fusion loses the SET methyltransferase domain is era nomenclature, not a factual conflict. In 1998-2002 usage the 'methyltransferase domain' or 'MT domain' of MLL denoted the CXXC region homologous to eukaryotic DNA methyltransferases: PMID 11842107 is titled 'The MT domain of the proto-oncoprotein MLL binds to CpG-containing DNA and discriminates against methylation' and defines it as 'A domain of MLL with significant homology to the eukaryotic DNA methyltransferases (MT domain)', which it further states 'has been found to be essential for the transforming potential of the oncogenic MLL derivatives'. PMID 12124344 independently describes 'a zinc-binding CXXC domain that MLL also contains within a methyltransferase domain', and PMID 9681821 lists 'the AT hooks and the DNA methyltransferase domain' among MLL's domains. Fetched coordinates confirm the structural reading: on ENST00000534358 / ENSP00000436786 (36 exons, 11,919 nt CDS, 3,972 aa), MLL exons 1-8 encode residues 1-1362, the CXXC domain at 1147-1190 falls in exon 5 and is retained, and the SET domain at 3829-3945 is encoded in exons 34-36 - 2,467 residues downstream of the exon-8 boundary and therefore absent from the chimera.

   The MT domain of the proto-oncoprotein MLL binds to CpG-containing DNA and discriminates against methylation. · Birke M, Schreiner S, García-Cuéllar MP, et al. · Nucleic Acids Research 2002
   PMID 11842107DOI
   ▸What this evidence cannot show

   THE EXON MAPPING MIXES TWO NUMBERING SOURCES. Domain coordinates are UniProt Q03164 (3,969 aa); exon boundaries and the CDS walk are Ensembl ENSP00000436786 (3,972 aa). The two differ by three residues, so this is not a single consistent coordinate system. The margin of the argument is 2,467 residues and no three-residue offset can move the SET domain from exon 36 into exon 8, but the mismatch is real and is stated rather than glossed. Separately, the nomenclature claim is established from three papers' usage, which is strong but is an argument about what authors meant rather than a statement any of them makes explicitly about the Daheron sentence.

   ▸Evidence that argues against this (2)

   • This resolution is an inference about authorial intent, and a weaker reading survives: Daheron 2001 might simply have been wrong about the fusion's domain content, in which case the graph's model would still stand but for a duller reason. The nomenclature argument is preferred because three independent papers from 1998-2002 use 'methyltransferase domain' for the CXXC region and because the intron-8 breakpoint makes the MT-domain reading structurally correct - but no paper states the equivalence for this sentence, and the entry should not be read as having quoted a source that does.
   • Retaining the MT/CXXC domain is not a neutral detail, and it argues against treating MLL-LPP as mechanistically boring. PMID 11842107 states the MT domain is 'essential for the transforming potential of the oncogenic MLL derivatives', so the domain the 2001 authors flagged is one with an asserted causal role - which means resolving the nomenclature flag does not dispose of the biology, and this graph has not tested whether the CXXC contribution differs between an intron-8 breakpoint and the intron-9-to-11 breakpoints typical of KMT2A::AFF1.
4. E4moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

   LPP's leukaemia literature is two primary papers, both from 2001, and the second was uncited in this graph. PMID 12063392 reports a novel LPP fusion transcript in a lipoma and states in its abstract that 'a fusion of LPP to the mixed lineage leukemia (MLL) gene has been described in a case of acute myeloid leukemia' - citing the same single case. A PubMed query for LPP[tiab] AND (MLL[tiab] OR KMT2A[tiab]) returns 3 records, of which the third is a 2026 paper on MEN1 variants of uncertain significance that contains no LPP token in any retrieval route. By contrast LPP's modern literature is a metastasis and mechanotransduction literature unconnected to the fusion: a Src substrate required for invadopodia formation and breast cancer lung metastasis (PMID 28436416), an inhibitor of collective cell migration in lung cancer dissemination (PMID 26028032), and a mechanotransducer in vascular smooth muscle cells (PMID 37759537).

   A novel LPP fusion gene indicates the crucial role of truncated LPP proteins in lipomas and pulmonary chondroid hamartomas. · Lemke I, Rogalla P, Bullerdiek J · Cytogenetics and Cell Genetics 2001
   PMID 12063392DOI
   ▸What this evidence cannot show

   'Two papers' is what a tiab query over MLL/KMT2A tokens finds, and is a LOWER BOUND, not a census: a case series reporting many partners in a table would not necessarily put LPP in its title or abstract, and cytogenetic literature from the 1990s indexes translocations by chromosome band rather than by gene. The claim is that LPP has essentially no dedicated fusion literature, not that no other paper has ever mentioned the rearrangement. The three modern papers are cited here from titles returned by a relevance-ranked sample; their abstracts were not read and no claim about their findings is made beyond what the titles state.

   ▸Evidence that argues against this (1)

   • The modern LPP literature is the reason to doubt this entry's framing rather than to extend it. Every one of those papers studies LPP in adherent solid-tumour or vascular contexts where mechanical tension exists and LPP is expressed; none is haematopoietic. Taking 'LPP is a mechanotransducer' into a leukaemia entry risks importing a function that the relevant cell type does not exercise, and the honest position is that the two literatures have no established point of contact.
5. E5moderate confidence 3 of 5: One strong study, or several smaller consistent ones.database

   A bare LPP symbol query is not usable. LPP[tiab] returns 4,215 records; the full name in either spelling returns 64. Hand-classification of a 60-record relevance-ranked sample of LPP[tiab] found only 10 that concern this gene, with the remainder spanning at least five unrelated senses - lichen planopilaris and lichen planus pigmentosus in dermatology, leak point pressure in urodynamics, the late positive potential in event-related-potential research, and lipopolysaccharide, lipopeptide and poly-lactic materials. Restricting to co-occurrence with the full name, with MLL or KMT2A, with HMGA2 or HMGIC (31), or with zyxin (25) is required for the token to be on-topic.

   NCBI E-utilities esearch/efetch against PubMed, 2026-09-22: LPP[tiab] n=4215 with 60 relevance-ranked records classified by hand; 'lipoma-preferred partner'[tiab] n=64; LPP[tiab] AND leukemia[tiab] n=12; LPP[tiab] AND (MLL[tiab] OR KMT2A[tiab]) n=3
   ▸What this evidence cannot show

   A 60-RECORD SAMPLE OF 4,215 IS 1.4%, AND IT WAS RELEVANCE-RANKED RATHER THAN RANDOM, so the 10-of-60 on-topic fraction is not an unbiased estimate of the corpus - relevance ranking against a gene-symbol query should if anything ENRICH for the gene, which makes 1-in-6 an optimistic ceiling rather than a point estimate. The sense classification was done by regex over title and abstract with a hand read of the titles, and 25 of 60 fell through as unclassified, so the five named senses are those that were recognised, not an exhaustive list.

   ▸Evidence that argues against this (1)

   • This graph's own collision register has a row recording that the approved symbol was the MINORITY token in 4 of 5 genes examined, and LPP inverts that: the symbol is 66 times more common than the full name. So neither 'prefer the symbol' nor 'prefer the full name' is a usable rule, which is an argument that per-gene measurement cannot be replaced by a heuristic - and therefore that the alias figures already stored for other genes in this graph do not generalise to genes whose splits were never measured.

• fuses with →KMT2A / MLL1 — the gene that breaks

  t(3;11)(q28;q23) in acute monoblastic leukaemia; breakpoint in intron 8 of both genes, MLL exon 8 to LPP exon 9, contributing LIM 2 (474-534) and LIM 3 (535-603). MLL exons 1-8 retain the CXXC/MT domain; the SET domain (exons 34-36) is lost.

• fuses with →HMGA2 (P52926) - a direct KMT2A::AFF1 target through the let-7b axis, and a recurrent fusion partner whose own fusions work by escaping a microRNA rather than by making a chimeric protein

  LPP is HMGA2's commonest partner in lipomas, pulmonary chondroid hamartomas and parosteal lipomas - the shared partner linking the two fusion families

• analogous to →KMT2A::AFF1 (MLL-AF4), t(4;11)(q21;q23)

  a KMT2A fusion partner from a different family: contributes two adhesion LIM domains rather than an elongation module, and is barely expressed in leukocytes