HOTTIP lncRNA — the 5' HOXA switch that holds the leukemic chromatin domain open
also called HOTTIP · HOXA distal transcript antisense RNA · HOXA transcript at the distal tip · HOXA13-AS
HOTTIP is a long RNA made from the far end of the HOXA gene cluster that does not code for a protein but instead helps switch the neighbouring HOXA genes on and keeps the DNA folded into the right loops. In KMT2A-rearranged and NPM1-mutant leukemia it is turned on abnormally, and turning it off slows leukemia in mice, which makes it one of the few non-coding elements in this disease with a real experimental case behind it.
The same thing, for a scientist
HOTTIP (ENSG00000243766; GRCh38 chr7:27,198,575-27,207,259, + strand, i.e. antisense to the minus-strand HOXA cluster) is a ~3.8 kb lncRNA transcribed from the 5' (distal, HOXA13-proximal) tip of HOXA. Mechanistically it binds the WD40 adaptor WDR5 directly and delivers WDR5/MLL complexes across HOXA to deposit H3K4me3, with chromosomal looping bringing HOTTIP into proximity with its targets [E1]. In leukemia it is a second, RNA-level input into the same HOXA program the KMT2A fusion drives: HOTTIP is aberrantly activated in AML driven by MLL rearrangement or NPM1 mutation, reshapes the HOXA topologically associated domain, and is required for leukemogenesis in transplanted mice, while forced Hottip expression in mice drives HSC self-renewal and an AML-like disease on its own [E2]. A later study from the same groups showed HOTTIP acts partly by nucleating R-loops at a subset of CTCF-binding sites, reinforcing CTCF boundary activity and TAD integrity, with the best-worked example being the beta-catenin TAD rather than HOXA itself [E3]. This is the strongest non-coding candidate in the KMT2A-r space, but its documented leukemia genetics are almost entirely AML/myeloid; the KMT2A::AFF1 infant B-ALL case is inferred, not demonstrated.
What the molecule actually does1 deeper
Two distinct activities have been demonstrated, and they are often conflated in reviews.
1. A WDR5/MLL delivery vehicle for H3K4me3. The founding paper identified HOTTIP as a lincRNA from the 5' tip of HOXA that coordinates activation of several 5' HOXA genes in vivo. HOTTIP RNA binds WDR5 directly and targets WDR5/MLL complexes across HOXA, driving H3K4 trimethylation. Critically, the paper showed that induced proximity is necessary and sufficient for HOTTIP activation of its targets — the RNA works in cis through the chromosomal loop, not as a diffusible factor 1.
This is the point of intersection with the project's central axis: HOTTIP recruits WDR5/MLL, the same WDR5-containing complex whose integrity matters for wild-type KMT2A activity. A KMT2A fusion protein has lost the SET domain and the WDR5-binding WIN motif region, so the fusion cannot use this route itself — but the wild-type allele and other MLL-family SET complexes can, which makes HOTTIP a candidate route by which the non-fusion allele reinforces the fusion's own target program.
2. An R-loop-dependent reinforcer of CTCF boundaries. HOTTIP directly interacts with and regulates a fraction of CTCF-binding sites in the AML genome, recruiting CTCF/cohesin and R-loop-associated regulators to form R-loops that reinforce the boundary and permit TAD formation 3. Deleting the boundary CBS, or targeting RNase H to that CBS to destroy the R-loop specifically, impaired boundary activity, cut promoter-enhancer interactions, reduced beta-catenin target expression and mitigated leukemogenesis in xenografts 3.
A correction worth recording: which paper is which
These two leukemia papers are routinely cited interchangeably and they are not the same result.
- Luo H et al., Cancer Cell 2019;36(6):645-659.e8, PMID 31786140 — "HOTTIP lncRNA Promotes Hematopoietic Stem Cell Self-Renewal Leading to AML-like Disease in Mice." This is the HSC self-renewal / transgenic mouse / HOXA TAD paper 2.
- Luo H et al., Molecular Cell 2022;82(4):833-851.e11, PMID 35180428 — "HOTTIP-dependent R-loop formation regulates CTCF boundary activity and TAD integrity in leukemia." This is Molecular Cell 2022, not Cancer Cell 2019 3.
The R-loop paper is frequently mis-cited as Cancer Cell 2019 in secondary literature. Both are from the Huang/Xu/Qiu/So collaboration, which is why they get merged.
The evidence tying HOTTIP to KMT2A-rearranged disease — and its limits
The honest statement is: HOTTIP is established as a HOXA-high AML lncRNA, and only inferred as a KMT2A::AFF1 B-ALL lncRNA.
What is demonstrated:
- HOTTIP is aberrantly activated in AML driven by MLL rearrangements or NPM1 mutations — the two canonical HOXA-high genotypes 2, 3.
- HOTTIP loss attenuates leukemogenesis of transplanted mice; re-expression restores leukemic TADs, transcription and leukemogenesis in CTCF-boundary-attenuated AML cells 2. This is a rescue experiment, which is the strongest form of the argument.
- Aberrant Hottip expression in mice abnormally promotes HSC self-renewal and produces AML-like disease 2. So it is not merely a passenger of HOXA activation; it is sufficient on its own in mouse.
What is not demonstrated:
- No published experiment isolates HOTTIP dependency in KMT2A::AFF1 infant B-ALL. The mouse work is myeloid, and the cell-line work is AML.
- The 2022 mechanism paper's decisive genetic experiments are at the beta-catenin TAD boundary, not at HOXA 3. The HOXA claim rests on correlative TAD/expression data. A reader who assumes the R-loop mechanism was proven at HOXA is over-reading the paper.
- HOTAIRM1 and miR-196b sit in the same locus and are co-regulated with the HOXA genes, so separating HOTTIP-specific effects from general HOXA-locus activation requires element-level perturbation that has not been done in a B-lymphoid KMT2A-r background.
Is this a plausible intervention point?
This is the most tractable non-coding target in the KMT2A-r space, for a specific technical reason: HOTTIP is a chromatin-associated nuclear lncRNA, and nuclear-retained transcripts are the class that RNase H1-recruiting gapmer ASOs degrade best, because RNase H1 is nuclear. siRNA/RISC, which is cytoplasmic, is the wrong tool. Published knockdowns used shRNA and locus-directed perturbation rather than ASOs, so ASO tractability is an inference from RNA localisation class, not a published result for this transcript.
Against it:
- HOTTIP's mechanism is cis-acting through induced proximity 1. Degrading the RNA should work; but the R-loop mechanism 3 means the functional entity is an RNA:DNA hybrid at a CTCF site, and it is unclear whether ASO-mediated cleavage of a chromatin-tethered, hybridised transcript is as efficient as cleavage of a free nuclear transcript.
- Redundancy: the same H3K4me3 output is reachable through wild-type KMT2A/WDR5 and through other MLL-family SET complexes, so removing the RNA delivery route may be buffered.
- Delivery to bone marrow is the real bottleneck. Systemic ASOs distribute well to liver and kidney and poorly to haematopoietic stem cells.
Verdict: ASO (RNase H gapmer) is the correct modality to test; no small molecule is plausible; the honest tractability grade is moderate, and the first useful experiment is not a drug but an ASO or CRISPRi knockdown in a KMT2A::AFF1 B-ALL PDX, which nobody has published.
Where this element sits relative to HOTAIRM1 and miR-196b
The HOXA cluster on GRCh38 chr7 packs at least four separately-annotated non-coding elements into ~110 kb:
| Element | GRCh38 chr7 | Strand | Position in cluster |
|---|---|---|---|
| HOTAIRM1 | 27,095,647-27,100,265 | + | 3' end, at HOXA1/HOXA2 |
| HOXA-AS2 | 27,107,777-27,134,302 | + | HOXA3-HOXA4 region |
| HOXA-AS3 | 27,129,977-27,155,928 | + | HOXA5-HOXA6 region |
| MIR196B | 27,169,480-27,169,563 | - | between HOXA9 and HOXA10 |
| HOTTIP | 27,198,575-27,207,259 | + | 5' tip, distal to HOXA13 |
Every one of them is on the opposite strand to the HOXA protein-coding genes except miR-196b. They are co-activated in KMT2A-r leukemia because the whole domain is activated; that is exactly why element-specific attribution is hard and why this beat is under-resolved. Coordinates retrieved from the Ensembl REST API, GRCh38 4.
| element symbol | HOTTIP |
|---|---|
| class | lncRNA |
| genome build | GRCh38 |
| locus | chr7:27,198,575-27,207,259 |
| chromosome | 7 |
| start | 27198575 |
| end | 27207259 |
| strand | + |
| cytoband | 7p15.2 |
| ensembl gene ID | ENSG00000243766 |
| ensembl canonical transcript | ENST00000521028 |
| hgnc ID | HGNC:37461 |
| position in cluster | 5' (distal) tip of the HOXA cluster, antisense, immediately adjacent to / overlapping HOXA13 (chr7:27,193,503-27,200,091, minus strand) |
| subcellular localisation | nuclear, chromatin-associated (acts in cis via induced proximity) |
| direction of change in kmt2a r | up |
| kmt2a r b all specific evidence | none published |
| functional dependency demonstrated in | mouse AML (transplantation, knockdown plus re-expression rescue); human AML xenograft |
protein partners
[ "WDR5", "MLL/KMT2A SET complex", "CTCF", "cohesin", "R-loop regulators" ]
cohorts with direction evidence
[
{
"cohort": "human AML cell lines and mouse MLL-fusion AML models",
"finding": "aberrantly activated in AML driven by MLL rearrangements or NPM1 mutations",
"pmid": "31786140",
"n": "not stated in abstract",
"title": "HOTTIP lncRNA Promotes Hematopoietic Stem Cell Self-Renewal Leading to AML-like Disease in Mice.",
"authors": "Luo H, Zhu G, Xu J, et al.",
"venue": "Cancer Cell",
"year": 2019,
"title_provenance": "Title/authors/venue backfilled from PubMed esummary on 2026-09-20 because this citation stored a bare PMID with no title, which made it unscoreable by verify-citations.py. BACKFILLING IS NOT VERIFICATION: it confirms the PMID resolves to a real paper, not that this is the right paper for the claim. That still requires reading."
}
]intervention assessment
{
"tractability": "moderate",
"best_modality": "RNase H1-recruiting gapmer antisense oligonucleotide (nuclear-retained lncRNA class), or CRISPRi at the HOTTIP promoter for target validation",
"modalities_ruled_out": [
"small molecule (no structured pocket; no published ligand)",
"siRNA/RISC (cytoplasmic machinery, wrong compartment for a chromatin-tethered transcript)"
],
"published_asos_against_this_target": "none found",
"key_uncertainty": "whether degrading a chromatin-tethered, R-loop-forming transcript is as efficient as degrading a free nuclear transcript, and whether bone-marrow delivery is achievable",
"first_experiment": "ASO or CRISPRi HOTTIP knockdown in KMT2A::AFF1 infant B-ALL PDX, with HOXA9/HOXA10 and beta-catenin-target readouts and a menin-inhibitor combination arm"
}- Redundancy: H3K4me3 at HOXA is reachable through wild-type KMT2A and other MLL-family SET complexes, so an ASO that removes the RNA delivery route may be fully buffered.
- Essential-function toxicity: HOTTIP's normal job is HOXA locus control in development and HSC self-renewal. An effective inhibitor should be expected to impair normal HSC self-renewal, which is precisely the reserve a relapsed patient does not have.
- Delivery: ASOs reach liver and kidney readily and bone-marrow HSCs poorly. A target that is only reachable with an ASO inherits an unsolved delivery problem.
- Compensation: HOTAIRM1, HOXA-AS2/AS3 and miR-196b are co-activated in the same domain. Silencing one element may be compensated within the locus, and the field has never tested element-by-element.
- Mechanism-linked resistance: if the functional entity is an RNA:DNA hybrid rather than free RNA, partial knockdown could leave enough chromatin-tethered transcript to maintain the boundary while looking like a successful knockdown by RT-qPCR of total RNA.
- Is HOTTIP a dependency in KMT2A::AFF1 B-lymphoid leukemia at all? Every functional experiment in the literature is myeloid. A single CRISPRi or ASO knockdown in SEM/RS4;11/MV4-11 plus an infant-ALL PDX would settle it and has not been done.
- Does the KMT2A fusion protein bind the HOTTIP locus directly, or is HOTTIP activation a downstream consequence of HOXA-domain opening? The published ChIP-seq datasets (Kerry 2017, Harman 2021) could answer this from existing data without a new experiment.
- Does menin inhibition switch HOTTIP off? If HOTTIP persists under revumenib while HOXA9 falls, HOTTIP is a candidate residual-disease mechanism and a combination target. If it falls with HOXA9 it is a passenger.
- Are HOTTIP-dependent R-loops present at the HOXA CTCF boundaries themselves, as the model implies but the 2022 paper did not show?
- Wild-type KMT2A is retained in every KMT2A-r patient. Does HOTTIP recruit the wild-type KMT2A/WDR5 complex to reinforce the fusion's own target program, making the untranslocated allele an accomplice?
- E1established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper
HOTTIP is a lincRNA transcribed from the 5' tip of the HOXA locus that coordinates activation of several 5' HOXA genes in vivo; HOTTIP RNA binds WDR5 directly and targets WDR5/MLL complexes across HOXA to drive H3K4 trimethylation, with induced proximity necessary and sufficient for target activation.
A long noncoding RNA maintains active chromatin to coordinate homeotic gene expression · Wang KC, Yang YW, Liu B, Sanyal A, Corces-Zimmerman R, Chen Y, Lajoie BR, Protacio A, Flynn RA, Gupta RA, Wysocka J, Lei M, Dekker J, Helms JA, Chang HY · Nature 472(7341):120-4 2011What this evidence cannot show
Developmental/fibroblast and limb-bud system, not leukemia. Knockdown was RNAi-based. The paper establishes the WDR5 mechanism but says nothing about KMT2A fusions. Later work in pancreatic cancer cells found HOTTIP did NOT regulate HOXA13 and that HOTTIP and HOTAIR control strikingly different gene sets, so the HOXA target set is cell-type dependent and should not be assumed from this paper.
Evidence that argues against this (1)
- In Panc1 pancreatic cancer cells HOTTIP knockdown did not regulate HOXA13 (contrary to the liver-cancer literature) and regulated HOXA10, HOXB2, HOXA11, HOXA9 and HOXA1 instead, showing the HOTTIP target set is not a fixed property of the locus. Cheng Y et al. 2015
- E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
HOTTIP is aberrantly activated in AML driven by MLL rearrangements or NPM1 mutations, altering the HOXA-driven TAD; HOTTIP loss attenuates leukemogenesis in transplanted mice and re-expression restores leukemic TADs, transcription and leukemogenesis in CTCF-boundary-attenuated AML cells, while Hottip aberration in mice promotes HSC self-renewal and causes AML-like disease.
HOTTIP lncRNA Promotes Hematopoietic Stem Cell Self-Renewal Leading to AML-like Disease in Mice · Luo H, Zhu G, Xu J, Lai Q, Yan B, Guo Y, Fung TK, Zeisig BB, Cui Y, Zha J, Cogle C, Wang F, Xu B, Yang FC, Li W, So CWE, Qiu Y, Xu M, Huang S · Cancer Cell 36(6):645-659.e8 2019What this evidence cannot show
Entirely myeloid: mouse AML models and human AML cells. No KMT2A::AFF1 B-ALL experiment. 'MLL rearrangements' in the abstract is not resolved to specific fusion partners, so the KMT2A::MLLT3-versus-KMT2A::AFF1 question cannot be answered from this paper. Transgenic overexpression of a lncRNA is a supraphysiological perturbation and the AML-like disease in mice does not establish that endogenous HOTTIP levels are rate-limiting in human disease.
Evidence that argues against this (1)
- No paper directly contradicts the HOTTIP-AML dependency, and that absence is itself informative: the finding has not been independently replicated by a group outside the Huang/Xu/Qiu/So collaboration, and the accompanying CRISPR-screen literature does not flag HOTTIP as a leukemia dependency, which would be expected if the effect size were large and general. Luo H et al. 2022
- E3moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper
HOTTIP directly interacts with and regulates a fraction of CTCF-binding sites in the AML genome by recruiting CTCF/cohesin and R-loop-associated regulators to form R-loops; deleting the boundary CBS or targeting RNase H to eliminate the R-loop at the beta-catenin TAD boundary impaired CTCF boundary activity, reduced promoter-enhancer interactions and beta-catenin target expression, and mitigated leukemogenesis in xenograft mice.
HOTTIP-dependent R-loop formation regulates CTCF boundary activity and TAD integrity in leukemia · Luo H, Zhu G, Eshelman MA, Fung TK, Lai Q, Wang F, Zeisig BB, Lesperance J, Ma X, Chen S, Cesari N, Cogle C, Chen B, Xu B, Yang FC, So CWE, Qiu Y, Xu M, Huang S · Molecular Cell 82(4):833-851.e11 2022What this evidence cannot show
Graded moderate rather than strong deliberately. The decisive loss-of-function genetics are at ONE locus (the beta-catenin TAD boundary), not at HOXA, so the generalisation to 'HOTTIP organises the leukemic genome' outruns the data. Targeted RNase H expression destroys all R-loops in a region and is not HOTTIP-specific. R-loop mapping is technically contested (DRIP-seq antibody artefacts are well documented) and the paper does not resolve whether the R-loop is cause or consequence of boundary occupancy. Same collaboration as E2, so not an independent replication.
Evidence that argues against this (1)
- No direct rebuttal published. The generic caveat is that R-loop detection by S9.6-antibody methods is prone to artefact, and no independent group has mapped HOTTIP-dependent R-loops. Luo H et al. 2019
- E4established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database
HOTTIP is ENSG00000243766, biotype lncRNA, GRCh38 chr7:27,198,575-27,207,259 on the plus strand, HGNC:37461, canonical transcript ENST00000521028; HOXA13 spans chr7:27,193,503-27,200,091 on the minus strand, so HOTTIP is transcribed antisense and distal to the 5' end of the HOXA cluster.
Ensembl REST API gene lookup, homo_sapiens, assembly GRCh38 (release 114) · Ensembl / EMBL-EBI · Ensembl 2026What this evidence cannot show
Gene-level coordinates of the Havana gene model; individual transcript isoforms differ and the frequently quoted 3,764 nt transcript length comes from a review rather than from the Ensembl record. lncRNA annotation boundaries are revised between Ensembl releases.
Evidence that argues against this (1)
- The same annotation caveat, and one specific to this locus: HOTTIP overlaps the HOXA13 gene model on the opposite strand, so read counts assigned to HOTTIP in unstranded or poorly stranded libraries can come from HOXA13 transcription. Expression claims about HOTTIP depend on library chemistry the primary reports do not always state. Ensembl, EMBL-EBI 2026
HOTTIP-dependent R-loops reinforce the CTCF boundary and RNase H targeting mitigates leukaemogenesis — R-loops here are load-bearing, not damage
- ← related toCTCF/cohesin boundaries and the HOXA topological domain — the 3D scaffold the fusion program runs on
Counter-evidence (Nora 2017, PMID 28525758): CTCF degradation abolishes insulation and CTCF-site looping while compartmentalisation is untouched, and restoring CTCF reinstates architecture - boundary activity and transcriptional output are separable read-outs, so the HOTTIP boundary rescue does not establish that architecture causes the transcriptional programme
5' HOXA lncRNA that holds the leukemic HOXA configuration open
- ← co-occurs withmiR-196b — the HOXA-cluster microRNA that is a direct KMT2A-fusion target and a genuine double agent
both are HOXA-cluster transcripts co-activated by the fusion; neither is independent of HOXA dosage
- supports →HOXA-high versus HOXA-low KMT2A-rearranged leukemia — the split that breaks the one-mechanism story
a candidate mechanism for how the HOXA-high state is held open rather than merely initiated
- regulates →Step 5 — The output program: HOXA9/HOXA10/MEIS1/PBX3 plus MEF2C, MYB, BCL2, FLT3 and PROM1/CD133
long non-coding RNA at the HOXA locus