DHX9 / RNA helicase A (Q08211) - the other TDRD3 effector, and it is itself the 3' partner of a leukaemia fusion

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.

• Nothing here is a target nomination. DHX9 has never been studied in a KMT2A-rearranged cell, its UniProt record spans five compartments and five broad processes, and the sole paper connecting it to TDRD3 is one record in the indexed literature.
• The therapeutic direction is unclear and probably unfavourable. DHX9 resolves R loops, so inhibiting it would be expected to RAISE transcription-replication conflict - and this project's own break-machinery beat is about how conflict at the KMT2A locus produces the rearrangements that cause this disease. Raising it in a child is a mutagenic proposition.
• The NPL-DHX9 fusion is a single case report with no functional characterisation and a single patient's cladribine response. It must not be cited as evidence that DHX9 fusions are drivers or that cladribine treats them.

1. Which effector carries the H4R3me2a-to-R-loop axis, DHX9 or TOP3B? Both are recruited by TDRD3, both are now written, and no experiment separates them. A double depletion with an R-loop readout at the same promoters is the obvious design and does not exist.
2. Why would TDRD3 deliver two anti-R-loop activities to the same chromatin? TOP3B prevents R loops by relaxing supercoiling and DHX9 resolves them by unwinding. Either they act at different loci or under different conditions, or one recruitment claim will not replicate.
3. Is the NDH2 alias contaminated as assumed? Its 191 records were not inspected. If a meaningful share are genuine DHX9 papers, this gene's literature is even more fragmented than the 6% Jaccard already indicates.
4. Does DHX9 sit at the KMT2A breakpoint cluster region? This graph's break-machinery beat argues the BCR is a site of transcription-replication conflict, DHX9 resolves exactly that, and DHX9 AND (MLL OR KMT2A) returns zero - so the question has never been asked.
5. Is the nucleolar pool separable from the nucleoplasmic one? UniProt lists both plus a centrosomal pool and a Ran-dependent shuttling mechanism, and this graph's subnuclear priority would be served by knowing which pool the TDRD3-recruited fraction comes from.

1. E1established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database

   DHX9 is UniProt Q08211, DHX9_HUMAN, ATP-dependent RNA helicase A, 1,270 aa, with synonyms DDX9, LKP and NDH2 and four annotated domains: DRBM 1 at 3-71, DRBM 2 at 180-252, a helicase ATP-binding domain at 398-564 and a helicase C-terminal domain at 636-809. UniProt describes it as a multifunctional ATP-dependent nucleic-acid helicase that unwinds both DNA and RNA in a 3' to 5' direction, with roles in DNA replication, transcriptional activation, post-transcriptional RNA regulation, mRNA translation and RNA-mediated gene silencing. Its subcellular location record lists five compartments - nucleus, nucleoplasm, nucleolus, cytoplasm and centrosome - and states that it is a nucleoplasmic shuttling protein whose nuclear import uses the importin-alpha/importin-beta pathway in a Ran-dependent manner.

   UniProtKB entry Q08211 (DHX9_HUMAN), retrieved by gene_exact:DHX9 AND organism_id:9606 AND reviewed:true with fields accession, id, protein_name, gene_names, length, cc_subcellular_location, ft_domain, cc_function, fetched 2026-09-22 (exactly one reviewed result)
   ▸What this evidence cannot show

   Accession, entry name, length, synonyms and all four domain coordinate pairs were read from the JSON response, and the query was an exact-gene search returning exactly one reviewed entry rather than a symbol guess. The FUNCTION comment is unusually broad and rests on roughly fourteen separate experimental citations, none of which was read here - so 'roles in DNA replication, transcription, translation and silencing' is a summary of a curated summary, and this entry makes no claim about which of those roles dominates in any cell type. The five-compartment localisation is likewise a union over experiments and cell types, not a statement about any one cell.

   ▸Evidence that argues against this (1)

   • A protein annotated for five compartments and five broad processes is a poor object for a specific mechanistic claim, and that cuts against the reason this entry was written. If DHX9 does everything everywhere, then 'TDRD3 recruits DHX9 to resolve R loops' describes one of many pools and cannot be assumed rate-limiting for the arginine-mark axis. The breadth is a warning, not a credential.
2. E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

   DHX9 is the second TDRD3 effector, and its role is R-loop resolution. TDRD3 - which recognises methylarginine marks on histones and on the C-terminal domain of RNA polymerase II - promotes DHX9 chromatin recruitment and R-loop resolution, with R loops particularly enriched at gene promoters where they regulate gene expression.

   TDRD3 promotes DHX9 chromatin recruitment and R-loop resolution. · Yuan W, et al. · Nucleic Acids Research 2021
   PMID 34329467DOI
   ▸What this evidence cannot show

   Abstract-level; the full text was not read, so the loci involved, the magnitude of the effect and whether DHX9 and TOP3B act at the same promoters are not established here. Not a haematopoietic or KMT2A-rearranged system. Verified live by efetch 2026-09-22. `DHX9[tiab] AND TDRD3[tiab]` returns exactly ONE record - this one - so the entire DHX9-TDRD3 relationship in the indexed literature rests on a single paper.

   ▸Evidence that argues against this (2)

   • This creates a problem for the entry that preceded it rather than solving one. gene-tdrd3 and gene-top3b both present the TDRD3-TOP3B route as the mechanism connecting H4R3me2a to R loops. With DHX9 in the same role from the same reader, there are two effectors and no published experiment separates them - so any claim that the axis runs through either one specifically is currently unsupported. The single-paper basis for the DHX9 arm means it is also the weaker-evidenced of the two.
   • The two effectors point in the same direction, which is itself odd and worth noting: TOP3B relaxes supercoiling to PREVENT R loops and DHX9 unwinds them to RESOLVE R loops, so TDRD3 would be delivering two redundant anti-R-loop activities to the same chromatin. Either they act at different loci or in different conditions, or one of the two recruitment claims will not survive replication.
3. E3weak confidence 2 of 5: Preliminary, small, indirect, or not yet peer reviewed.paper

   DHX9 is itself the 3' partner of a leukaemia fusion gene. A 30-year-old patient with T-cell large granular lymphocytic leukaemia - a rare, usually indolent lymphoproliferative disorder - carried an NPL-DHX9 gene fusion and was treated successfully with cladribine. Separately, DHX9 has been implicated in myelodysplastic syndromes: expression was analysed in 120 MDS patients against 42 non-MDS controls, with lentivirus-mediated DHX9-knockdown experiments following up.

   Case Report: First Report of T-Cell Large Granular Lymphocytic Leukemia With NPL-DHX9 Gene Fusion Successfully Treated With Cladribine: Clinical Experience and Literature Review. · Hu N, et al. · Frontiers in Oncology 2023
   PMID 35600388DOI
   ▸What this evidence cannot show

   A SINGLE CASE REPORT, self-described as a first report, in a disease with no relationship to KMT2A-rearranged infant ALL beyond both being leukaemias. Abstract-level only: whether the fusion was shown to be functional, in-frame, or expressed as a protein is not established here, and a case report cannot establish that it was the driver rather than an incidental finding. The cladribine response is one patient. The MDS work (PMID 37305700, Huang et al., iScience 2023, doi 10.1016/j.isci.2023.106962, verified separately) is a cohort plus knockdown and is stronger, but it is MDS rather than a fusion-driven leukaemia. Both PMIDs were uncited in this graph before this entry.

   ▸Evidence that argues against this (1)

   • Including a single case report under the PI's analogous-fusion-families priority risks inflating that dimension with noise. A fusion seen once, in an indolent T-cell disorder, whose functional consequence is unreported, is not comparable to the NUP98, EWSR1-FLI1 or SFPQ-ABL1 comparators this graph carries - each of which has mechanism behind it. It is recorded here because the PI asked for analogous fusion biology and because a partner protein that is itself a fusion partner is a fact worth being able to find, not because it supports an inference.
4. E4strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.database

   The approved symbol and the protein name retrieve almost disjoint literatures. DHX9[tiab] returns 344 records and "RNA helicase A"[tiab] returns 223, and their intersection is 32 - a Jaccard index of 6.0%. DDX9[tiab] returns 3. The listed synonym NDH2[tiab] returns 191 but is not counted as DHX9 literature because NDH2 also names alternative/type-II NADH dehydrogenases in fungi, plants and bacteria. DHX9[tiab] AND (MLL[tiab] OR KMT2A[tiab]) returns 0 with a parsed translation and warninglist ['No items found.'].

   NCBI esearch against PubMed with ID sets compared at retmax=500, 2026-09-22; querytranslation and warninglist recorded for the zero.
   ▸What this evidence cannot show

   ID-set comparison at retmax=500 is complete for these counts (the largest set is 344) so nothing was truncated. The NDH2 exclusion is a judgement, not a measurement: its 191 records were not individually inspected, so the true DHX9 share of them is unknown and could be non-trivial - which would make the alias problem worse rather than better. The KMT2A zero bounds indexed titles and abstracts only and was not re-run a second time, so it is reported as moderate-strength within a strong-confidence item about the alias split.

   ▸Evidence that argues against this (1)

   • A 6% Jaccard index between a gene symbol and its own protein name could reflect two literatures about genuinely different things rather than one literature split by naming - older biochemistry on 'RNA helicase A' as a purified activity versus modern genomics on 'DHX9' as a gene. If so, unioning them indiscriminately would import non-equivalent evidence, and the right response is to read both sets rather than simply to add them.

• ← analogous toNUMA1 / NuMA (Q14980) - a nuclear-matrix protein that becomes the spindle pole, and the 5' partner of a variant APL fusion whose coiled-coil is required for oncogenicity

  same shape, different evidence depth: DHX9's fusion is one T-LGLL case report, NuMA's is a Nature Genetics founding paper plus a required-domain result plus transgenic mice

• ← recruitsTDRD3 (Q9H7E2) - the reader the new H4R3me2a entry could not name, and it brings TOP3B to the MYC promoter and keeps R loops off it

  TDRD3 promotes DHX9 chromatin recruitment and R-loop resolution (PMID 34329467) - the SECOND effector, alongside TOP3B; gene-tdrd3 presents TOP3B as the mechanism and does not mention this one

• co-occurs with →TOP3B (O95985) - the only human topoisomerase that works on RNA as well as DNA, delivered to the MYC promoter by an arginine mark, and it has poisons

  the two TDRD3 effectors, delivered by one reader to do redundant anti-R-loop work: TOP3B relaxes supercoiling to prevent R loops, DHX9 unwinds to resolve them; no experiment separates them

• related to →H4R3me2a - the activating arginine mark that a KMT2-family enzyme's OWN tandem PHD module reads, where the Kabuki-syndrome mutations sit, and which no KMT2A-fusion study has ever measured

  indirect and mark-dependent only through the reader: DHX9 has no arginine-mark contact of its own, it is recruited by TDRD3 which reads the mark

• related to →Transcription–replication conflict as the candidate reason KMT2A-rearranged cells need the Fanconi pathway — a chain with two measured links and three inferred ones

  DHX9 resolves exactly the conflict this mechanism invokes, and DHX9[tiab] AND (MLL OR KMT2A) returns 0 - the question has never been asked

• tests →Measure replication stress directly in KMT2A-rearranged cells instead of inferring it from a dependency ranking

  a DHX9 arm would belong in this readout; its absence from the design is unremarked because no DHX9 entry existed until cycle 57

• analogous to →SFPQ / PSF (P23246) - the last of 21 fusion-construct partners without an entry, and it is itself the 5' half of an ALL fusion that puts ABL1 in the nucleus

  both are interactome-layer RNA-binding proteins that are ALSO fusion partners in their own right - SFPQ-ABL1 in B-ALL, NPL-DHX9 in T-LGLL (PMID 35600388); DHX9 is the second instance of the shape in this graph

• ← bindsNUP98 - the mobile nucleoporin that leaves the pore to put H3K4me3 on haematopoietic promoters, and the shared 5' donor of seven fusions in this graph

  intranuclear NUP98 binds DHX9 through its FG/GLFG region in an RNA-facilitated manner and stimulates its ATPase