Human mRNA nuclear export — TREX loading on spliced mRNA, NXF1-NXT1 handover, nuclear pore translocation and DDX19-GLE1 remodelling, with the CRM1 route for specific mRNAs; THOC1/THOC2/THOC3/THOC5/THOC6/THOC7/DDX39B/ALYREF/CHTOP/SARNP/NXF1/NXT1/MCM3AP/ENY2/PCID2/DDX19B/GLE1/NUP214/XPO1/RAN

A mature mRNA, capped, spliced and polyadenylated in the nucleus, must cross the nuclear pore complex (NPC) before it can be translated. In human cells bulk export is coupled to splicing and to the cap. The transcription-export complex TREX, built from the multi-subunit THO complex (THOC1, THOC2, THOC3, THOC5, THOC6, THOC7) with the DEAD-box ATPase UAP56 (DDX39B) and the export adaptors ALYREF, CHTOP and CIP29 (SARNP), is deposited near the 5' end of the spliced mRNA through an interaction between ALYREF and the cap-binding protein CBP80. UAP56 uses its ATPase cycle to load ALYREF and CHTOP onto the mRNA. The adaptors then recruit the export receptor, the NXF1-NXT1 (TAP-p15) heterodimer: ALYREF and THOC5 contact NXF1 and switch it from an autoinhibited state into one that exposes its arginine-rich RNA-binding domain, and the mRNA is handed over from ALYREF to NXF1. TREX-2 (GANP/MCM3AP with ENY2 and PCID2), anchored on the nuclear basket, helps deliver NXF1-bound mRNPs to the pore. NXF1-NXT1 carries the mRNP through the central channel by transient contacts between its NTF2-like and UBA domains and the phenylalanine-glycine (FG) repeats of the nucleoporins, on a scaffold whose outer rings are built from the Nup107-160 (Y) complex. At the cytoplasmic filaments the DEAD-box ATPase DDX19 (DDX19B), anchored on NUP214, is activated by GLE1 and its small-molecule cofactor inositol hexakisphosphate (IP6). DDX19 remodels the emerging mRNP, stripping export factors so that they recycle to the nucleus and making the passage directional. A minority of mRNAs leave by a second, Ran-dependent route through the karyopherin CRM1 (XPO1), which is recruited by mRNA-specific adaptors: LRPPRC with eIF4E for mRNAs that carry the eIF4E-sensitivity element, HuR (ELAVL1) with pp32 and APRIL (ANP32A, ANP32B) for AU-rich-element mRNAs, NXF3, and FMRP for N6-methyladenosine-marked mRNAs. RanGTP loads the cargo onto CRM1 in the nucleus, and GTP hydrolysis by Ran, stimulated by RanGAP1 with RanBP1 and RanBP2 at the cytoplasmic face, releases it.

Derived from the RGD Pathway Ontology term PW:0001587 (mRNA nuclear export pathway) through the RGD_PATHWAY_ONTOLOGY project's diagram->module route. RGD curates 37 human genes to it. All 37 are modelled: the TREX subunits ALYREF, DDX39B, SARNP and THOC1/2/5/6/7; NXF1 and NXT1; the nucleoporins NUP62, NUP54, NUP58, NUP98, NUP153 and NUP214 and the Y-complex members NUP107, NUP133, NUP160, NUP85, NUP37, NUP43, SEC13 and SEH1L; DDX19B and GLE1; and the CRM1 branch XPO1, RAN, RANGAP1, RANBP1, RANBP2, LRPPRC, EIF4E, ELAVL1, ANP32A, ANP32B and NXF3. Added beyond the RGD list, each with its own evidence: THOC3 (the Tex1 subunit, present in the fly module and in TREX by GO's definition), CHTOP, POLDIP3 and ZC3H11A (TREX adaptors shown biochemically to join human TREX), the TREX-2 subunits MCM3AP (GANP), ENY2 and PCID2, and FMR1 (from the human GO-CAM). The mapped GO process GO:0006406 is the module concept. Of the diagram's sub-pathway links, the NXF1-NXT1 export pathway (PW:0001625) is fully modelled and listed as evidence; the CRM1 export pathway is modelled only for its mRNA branch, because CRM1 also exports proteins, snRNA and ribosomal subunits, so that RGD term is mentioned here but not claimed. The spliceosome, 5' capping and 3'-end processing sub-pathways are upstream inputs that produce the mature mRNA and are outside the boundary. Fly counterpart. modules/dmel_mrna_nuclear_export.yaml models the conserved core from the FlyBase TREX, THO-C, TREX-2 and NXF gene groups: TREX (Hel25E, Ref1, THO, tex), the Nxf1-Nxt1 receptor (sbr, Nxt1) and TREX-2 (xmas, PCID2, e(y)2), and puts the NPC and the EJC outside. This module keeps those three parts in the same order and with the same complex terms, so the two can be compared directly, and adds what the RGD pathway adds: the NPC translocation step, the DDX19-GLE1 remodelling step and the CRM1 route. Two small differences: TREX-2 is marked optional here (in human cells it behaves like a basket component and its depletion slows export rather than abolishing it, PMID:23591820), and the NXF1-NXT1 complex is grounded on GO:0042272 nuclear RNA export factor complex, whose exact synonyms are "NXF1-NXT1 complex" and "TAP-p15 complex". The fly module states that GO has no CC term for this heterodimer; GO:0042272 is that term and could be used there. The fly module is referenced in prose only, since a cross-species module_ref would assert an identity this module does not need. Nuclear pore. No human nuclear pore complex module exists in modules/ (only dmel_nuclear_pore_complex), so module_ref was not used and the pore is modelled inline as a required transport step with two complex annotons: the FG-nucleoporins the receptor contacts (NUP62, NUP54, NUP58, NUP98, NUP153) and the Y-complex scaffold. They are parts, not context, because RGD includes them in the pathway and the receptor-FG contact is the translocation mechanism, but the step is a shared transport organelle used by every karyopherin route, and if a human NPC module is written later this step should become a module_ref to it. The Y-complex annoton carries no mRNA-export process: it is the scaffold the FG layer sits on, and its contribution to export is structural. NUP96, which is produced from the same NUP98 precursor and belongs to the Y-complex, shares NUP98's accession and is represented only through it. TPR and NUP50 of the basket, NUP88 and NUP42 (hCG1, the GLE1 anchor) of the cytoplasmic face and the other nucleoporins are not listed; NUP42 is omitted because no cached human evidence for its GLE1 role was available. Boundary decisions. (1) TREX-2 is an optional part rather than context, following the fly module and the GANP evidence that it binds NXF1 and is needed for efficient NXF1 recruitment to pores. Its centrin (CETN2/CETN3) and DSS1 (SEM1) subunits, known from yeast and human biochemistry, are not listed. (2) The exon junction complex is context, not a part, as in the fly module. It is deposited by splicing and some of its components help recruit TREX, but its core function is in mRNA localisation, translation and nonsense-mediated decay, and RGD does not include it. The cap-binding complex (CBP80/NCBP1), which recruits TREX through ALYREF, and the SR-protein export adaptors (SRSF3, SRSF7) that also hand mRNA to NXF1 are likewise context, recorded in prose. (3) The CRM1 route is included as an optional part because it is half of the RGD pathway and is the route of the only production human mRNA export GO-CAM. Its adaptors form a ROUTE variant set, since they define different mRNA classes, not interchangeable carriers of one activity. RAN, RANGAP1, RANBP1 and RANBP2 are modelled inside the CRM1 part as the Ran cycle that drives cargo loading and release; the RanGDP import and RCC1 recharging steps of the Ran cycle are outside. NUP98's role in the CRM1 routes is noted on its unit rather than duplicated. (4) DDX39A (URH49), the paralog of UAP56, can substitute for it in TREX-like complexes, and DDX19A is a paralog of DDX19B; neither is in the RGD list and neither is modelled. The one human GO-CAM containing DDX39A (gomodel:66c7d41500002513) models it as a negative regulator of immune-related RNA export and is not linked here. (5) Inositol hexakisphosphate is named in prose and in GLE1's function (GO:0000822) rather than as a separate participant; its synthesis by IPMK and IPPK is outside the boundary. Molecular functions were taken from each protein's GO annotations in QuickGO (checked November 2026), preferring experimental rows. DDX39B GO:0003724 RNA helicase activity (IDA); ALYREF GO:0003729 mRNA binding (IBA, with direct binding shown in PMID:18364396); THOC2 GO:0003729 (IBA); THOC5 GO:0030674 protein-macromolecule adaptor activity (IBA); NXF1 GO:0003729 (its GOA rows are GO:0003723 RNA binding by HDA/IBA; the mRNA-specific child is used on the strength of PMID:18364396); nucleoporins GO:0017056 structural constituent of nuclear pore (IDA/IMP for NUP107, NUP133, NUP160, NUP153, NUP98; IBA for the rest); DDX19B GO:0003724 (IBA; GO:0016887 ATP hydrolysis activity by EXP); GLE1 GO:0000822 inositol hexakisphosphate binding (IBA); XPO1 GO:0005049 nuclear export signal receptor activity (IDA/IMP); RAN GO:0003924 GTPase activity (IDA); RANGAP1 GO:0005096 GTPase activator activity (IDA); EIF4E GO:0000340 RNA 7-methylguanosine cap binding (IDA); LRPPRC GO:0003729 (IDA; core in its gene review); ELAVL1 GO:0035925 mRNA 3'-UTR AU-rich region binding (IDA; core in its gene review); NXF3 GO:0003723 RNA binding (IBA); FMR1 GO:1990247 (from the GO-CAM). Units with no GO function that describes their role here (CHTOP, SARNP, POLDIP3, ZC3H11A, THOC1/3/6/7, NXT1, MCM3AP, ENY2, PCID2, NUP37, NUP43, SEC13, SEH1L, RANBP1, RANBP2, ANP32A, ANP32B) carry either no function or a free-text preferred_term, rather than an off-role GO term such as CHTOP's methyl-CpG binding or ANP32B's histone binding. Bare protein binding is not used. GO has no molecular-function term for an mRNA export receptor or export adaptor (GO:0140142 nucleocytoplasmic carrier activity is do-not-annotate, and GO:0005049 is defined by the protein NES), which is why NXF1 and ALYREF carry mRNA binding; this is an ontology gap, not a biological one. GO-CAM and GOA observations. gocams/index.tsv (November 2026) has one production human model of mRNA export, gomodel:63894f2500001748 (FMRP-XPO1, m6A mRNAs), linked from the XPO1 and FMR1 annotons, and its mouse counterpart gomodel:63894f2500001789. The seed's best GO-CAM lead is that same model. No production GO-CAM models TREX, NXF1-NXT1, TREX-2, the NPC step or DDX19-GLE1, which is the largest curation gap the module reveals. RAN's only GO-CAM (gomodel:61f34dd300000863) and RANBP2's (gomodel:66c7d41500002513) model protein export and SUMOylation respectively and are not linked. Two possible upstream issues: GLE1 has no annotation to GO:0001671 ATPase activator activity although activation of DDX19 is its defining activity (yeast Gle1 evidence; human data would be needed for a direct annotation), and NUP214 and NUP42 carry GO:0005049 nuclear export signal receptor activity by IDA/IBA, a karyopherin term that does not obviously describe a nucleoporin and is worth an upstream look. Gene reviews exist only for NUP98, LRPPRC and ELAVL1; LRPPRC's review treats its nuclear export role as non-core next to its mitochondrial function, which this module respects by placing it only in an optional route. PANTHER family ids and PTNs are not asserted.