NSUN2

UniProt ID: Q08J23
Organism: Homo sapiens
Review Status: COMPLETE
📝 Provide Detailed Feedback

Gene Description

RNA cytosine-5 methyltransferase, the human orthologue of yeast Trm4, and the enzyme responsible for most 5-methylcytosine in human RNA. NSUN2 uses S-adenosyl-L-methionine and a catalytic cysteine to methylate cytosines in transfer RNA: the wobble base C34 of intron-containing pre-tRNA-Leu(CAA), and C47-C50 in the variable loop of many cytoplasmic tRNAs. It is imported into the mitochondrial matrix, where it makes m5C at positions 48-50 of several mitochondrial tRNAs, and it is the sole enzyme that methylates C69 of the vault RNA VTRNA1.1, thereby directing its processing into small vault RNAs. The methylation it deposits protects tRNAs from stress-induced endonucleolytic cleavage, and loss of the enzyme alters translation, tRNA-fragment production and the differentiation of skin, hair follicle and neural stem cells; biallelic loss-of-function variants cause autosomal-recessive intellectual disability and a Dubowitz-like syndrome. NSUN2 is predominantly nuclear and nucleolar, with cytoplasmic and mitochondrial pools, and it has been detected in secreted exosomes. NSUN2 also methylates messenger RNA, and m5C in mRNA has been linked to nuclear export via ALYREF and to transcript stabilisation via YBX1; this substrate class is genuinely contested, because the mapping methods used to locate mRNA m5C sites - bisulfite sequencing, Aza-IP and miCLIP - agree poorly with one another, mRNA structure generates false-positive bisulfite signals, and the modification occurs at roughly one in several thousand cytosines, which leaves it unclear how the proposed reader proteins discriminate modified transcripts.

Existing Annotations Review

GO Term Evidence Action Reason
GO:0009383 rRNA (cytosine-C5-)-methyltransferase activity
IBA
GO_REF:0000033
MARK AS OVER ANNOTATED
Summary: Phylogenetic transfer of rRNA C5 methyltransferase activity across the NSUN family node.
Reason: The only donor named in the GOA WITH/FROM column besides the family node is mouse Nsun4 (MGI:MGI:1919431), whose characterised substrate is mitochondrial 12S rRNA. NSUN2 has a well-mapped and quite different substrate repertoire - cytoplasmic pre-tRNA C34, tRNA and mt-tRNA C47-C50, vault RNA C69, and mRNA - and no rRNA substrate has been reported for it in twenty years of work on the enzyme. This looks like a paralogue-specific function propagated too broadly rather than a genuine ancestral activity retained by NSUN2. Not removed outright, since the PAINT tree itself was not inspected and an ancestral rRNA-methylating activity for the family is plausible; marked as over-annotation of this particular gene.
Propagation Review
Root cause: PROPAGATION BAD
Failure modes: WRONG ORTHOLOG OR PARALOG FUNCTIONAL DIVERGENCE
Sources checked:
MGI:MGI:1919431 · Nsun4 SUPPORTS SOURCE BUT NOT TARGET
Donor named in the GOA WITH/FROM column is mouse Nsun4, the mitochondrial 12S rRNA m5C methyltransferase that partners MTERF4 in mitoribosomal large subunit assembly. That is a paralogue-specific substrate and complex; NSUN2 has no reported rRNA substrate.
PANTHER:PTN000516076 · PTN000516076 UNRESOLVED
Family-wide NSUN node. The PAINT tree was not inspected for this review, so no claim is made about where the IBD was placed; the objection rests only on the donor gene identity and on NSUN2-specific substrate data.
GO:0016428 tRNA (cytidine-N5)-methyltransferase activity
IBA
GO_REF:0000033
ACCEPT
Summary: Phylogenetic assertion that NSUN2 inherits the Trm4-type tRNA m5C methyltransferase activity of its family.
Reason: Correct and core. The donor list spans mouse, fission yeast (two genes), budding yeast and C. elegans as well as NSUN2 itself, which is the expected pattern for a deeply conserved function: NSUN2 appearing in its own WITH/FROM simply reflects that its experimental annotation was one of the descendant evidences used to place the ancestral node, and the IBA adds the further statement that the activity is inherited rather than human-specific.
Supporting Evidence:
PMID:17071714
We identified a human orthologue of tRNA:m5C methyltransferase from Saccharomyces cerevisiae, which has been previously shown to catalyse the specific modification of C34 in the intron-containing yeast pre-tRNA Leu (CAA).
GO:0002127 tRNA wobble base cytosine methylation
IBA
GO_REF:0000033
ACCEPT
Summary: Methylation of the wobble cytosine C34 of tRNA, inherited from the Trm4 ancestor.
Reason: Directly matched by human experimental work: m5C34 is introduced into intron-containing pre-tRNA-Leu(CAA) in HeLa extract, and the human enzyme complements a yeast trm4 deletion.
Supporting Evidence:
PMID:17071714
Using transcripts of intron-less and intron-containing human pre-tRNA Leu (CAA) genes as substrates, we have shown that m5C34 is introduced only in the intron-containing tRNA precursors when the substrates were incubated in the HeLa extract.
GO:0005634 nucleus
IBA
GO_REF:0000033
ACCEPT
Summary: NSUN2 is active in the nucleus, where pre-tRNA methylation occurs.
Reason: Correct core location; NSUN2 is predominantly nuclear, with nucleolar concentration.
Supporting Evidence:
PMID:31287866
Although NSUN2 is predominantly localized to the nucleus and introduces m5C into cytoplasmic tRNAs and mRNAs, structured illumination microscopy clearly revealed NSUN2 foci inside mitochondria.
GO:0005737 cytoplasm
IBA
GO_REF:0000033
ACCEPT
Summary: NSUN2 is also active in the cytoplasm.
Reason: Correct; a cytoplasmic pool is seen by immunofluorescence and NSUN2 methylates cytoplasmic tRNAs.
GO:0006364 rRNA processing
IBA
GO_REF:0000033
MARK AS OVER ANNOTATED
Summary: rRNA processing propagated across the NSUN family node.
Reason: Same propagation issue as GO:0009383: the named donor is mouse Nsun4, whose role is in mitochondrial ribosome biogenesis. NSUN2 is nucleolar, so a role in ribosome biogenesis is not absurd, but no NSUN2-dependent rRNA processing defect has been demonstrated, and the assertion here is inherited from a paralogue rather than grounded in NSUN2 data.
Propagation Review
Root cause: PROPAGATION BAD
Failure modes: WRONG ORTHOLOG OR PARALOG FUNCTIONAL DIVERGENCE
Sources checked:
MGI:MGI:1919431 · Nsun4 SUPPORTS SOURCE BUT NOT TARGET
Donor named in the GOA WITH/FROM column is mouse Nsun4, the mitochondrial 12S rRNA m5C methyltransferase that partners MTERF4 in mitoribosomal large subunit assembly. That is a paralogue-specific substrate and complex; NSUN2 has no reported rRNA substrate.
PANTHER:PTN000516076 · PTN000516076 UNRESOLVED
Family-wide NSUN node. The PAINT tree was not inspected for this review, so no claim is made about where the IBD was placed; the objection rests only on the donor gene identity and on NSUN2-specific substrate data.
GO:1902775 mitochondrial large ribosomal subunit assembly
IBA
GO_REF:0000033
MARK AS OVER ANNOTATED
Summary: Mitochondrial large ribosomal subunit assembly propagated across the NSUN family node.
Reason: This is the documented function of NSUN4, which forms a complex with MTERF4 and is required for mitoribosome large-subunit assembly; the GOA WITH/FROM column names mouse Nsun4 as the donor. NSUN2 is genuinely imported into the mitochondrial matrix, but its demonstrated job there is m5C at positions 48-50 of mt-tRNAs, and its loss does not have a profound effect on mitochondrial translation or oxidative phosphorylation in differentiated cells (PMID:31276587) - which is what one would expect to be perturbed if NSUN2 were needed for large-subunit assembly.
Propagation Review
Root cause: PROPAGATION BAD
Failure modes: WRONG ORTHOLOG OR PARALOG FUNCTIONAL DIVERGENCE
Sources checked:
MGI:MGI:1919431 · Nsun4 SUPPORTS SOURCE BUT NOT TARGET
Donor named in the GOA WITH/FROM column is mouse Nsun4, the mitochondrial 12S rRNA m5C methyltransferase that partners MTERF4 in mitoribosomal large subunit assembly. That is a paralogue-specific substrate and complex; NSUN2 has no reported rRNA substrate.
PANTHER:PTN000516076 · PTN000516076 UNRESOLVED
Family-wide NSUN node. The PAINT tree was not inspected for this review, so no claim is made about where the IBD was placed; the objection rests only on the donor gene identity and on NSUN2-specific substrate data.
Supporting Evidence:
PMID:31276587
Finally, we show that inactivation of NSUN2 does not have a profound effect on mitochondrial tRNA stability and oxidative phosphorylation in differentiated cells.
GO:0062152 mRNA (cytidine-5-)-methyltransferase activity
IBA
GO_REF:0000033
KEEP AS NON CORE
Summary: Phylogenetic assertion of mRNA m5C methyltransferase activity, from a narrower node (PTN002736915) than the tRNA assertion.
Reason: GOA carries both sides of this question and both are retained. The catalytic assignment itself is supported by evidence that does not depend on the contested mapping assays: RNA-mediated activity-based protein profiling traps NSUN2 covalently on 5-fluorocytidine-containing RNA recovered by oligo-dT capture, and nucleoside LC-MS/MS of polyA-selected RNA shows an 82% fall in m5C in NSUN2 knockout cells (PMID:34556860); the mRNA export phenotype is rescued by wild-type but not methyltransferase-dead NSUN2 (PMID:28418038); and in vitro methylation of a defined p16 3'UTR fragment was reported (PMID:22395603). The activity is therefore not removed - and could not be, since this term also carries an EXP code whose full text has not been read here. What is disputed is the reliability and biological weight of the mRNA m5C maps. The methods disagree with one another (very low overlap between Aza-IP, bisulfite sequencing and miCLIP), mRNA secondary structure shields cytosines from bisulfite and generates false positives, and m5C occurs at roughly one in several thousand cytosines in mRNA, which leaves it unexplained how the proposed readers discriminate modified transcripts (PMID:42587746). The authors of the chemoproteomic study themselves note that polyA fractions carry ~40-fold lower m5C than total RNA and that contaminating tRNA/rRNA nucleosides could contribute to the measured signal. Demoted to non-core rather than accepted as core: the deeply conserved, structurally and genetically nailed-down substrate class for this Trm4 orthologue is tRNA (and vault RNA and mitochondrial tRNA), whereas the mRNA activity is quantitatively minor, its site maps are unresolved, and the large disease literature built on single mRNA m5C sites rarely supplies orthogonal validation.
GO:0005576 extracellular region
IEA
GO_REF:0000044
MARK AS OVER ANNOTATED
Summary: Extracellular region, mapped from the UniProt subcellular location line that records detection in secreted exosomes.
Reason: Reads as though a nuclear and cytoplasmic RNA methyltransferase were a secreted protein. The underlying observation - NSUN2 in exosomes from HEK293 cells - is captured more precisely by GO:0070062 extracellular exosome, which this gene also carries.
GO:0005730 nucleolus
IEA
GO_REF:0000044
ACCEPT
Summary: Nucleolar localisation mapped from the UniProt subcellular location vocabulary.
Reason: Consistent with the direct immunofluorescence localisation of the human enzyme.
GO:0005737 cytoplasm
IEA
GO_REF:0000120
ACCEPT
Summary: Cytoplasmic localisation from combined automatic methods.
Reason: Correct; agrees with the IDA and EXP cytoplasm annotations on this gene.
GO:0005739 mitochondrion
IEA
GO_REF:0000044
ACCEPT
Summary: Mitochondrial localisation from the UniProt subcellular location vocabulary.
Reason: Supported by two independent 2019 studies using proximity labelling and structured illumination microscopy, and by the mt-tRNA substrate.
GO:0005819 spindle
IEA
GO_REF:0000044
KEEP AS NON CORE
Summary: Spindle localisation from the UniProt subcellular location vocabulary.
Reason: NSUN2 has a reported mitotic spindle association, but this is not where the characterised RNA methylation chemistry happens and no spindle-associated substrate has been defined. Retained as a peripheral localisation.
GO:0006397 mRNA processing
IEA
GO_REF:0000117
MODIFY
Summary: GO:0006397 mRNA processing covers conversion of a primary transcript into mature mRNA - splicing, capping, polyadenylation. The cited work reports mRNA stabilisation and nuclear export instead.
Reason: A term-choice problem rather than an evidence problem, so the annotations are modified rather than removed. PMID:22395603 reports that NSUN2 inhibits turnover of p16 mRNA, PMID:31358969 that NSUN2 with YBX1 stabilises HDGF mRNA, and PMID:28418038 that NSUN2-dependent m5C promotes ALYREF-mediated nuclear export. None of these is a pre-mRNA maturation step. The accurate terms are mRNA stabilization (GO:0048255) and regulation of mRNA export from nucleus (GO:0010793), the latter of which this gene already carries from PMID:28418038.
GO:0008168 methyltransferase activity
IEA
GO_REF:0000002
MODIFY
Summary: Generic methyltransferase activity from the InterPro family signature.
Reason: Correct but far too general for a gene whose exact reaction, substrate and modified positions are established. The specific child is already annotated from experiment.
GO:0008173 RNA methyltransferase activity
IEA
GO_REF:0000002
MODIFY
Summary: RNA methyltransferase activity from the InterPro family signature.
Reason: Correct but under-specific; replaced by the established reaction-level child. Note that for NSUN2, unlike its pseudoenzyme paralogue NSUN7, the family signature does correspond to a real catalytic activity.
GO:0016428 tRNA (cytidine-N5)-methyltransferase activity
IEA
GO_REF:0000120
ACCEPT
Summary: tRNA m5C methyltransferase activity from combined automatic methods.
Reason: Agrees with the experimental annotations; core molecular function.
GO:0030488 tRNA methylation
IEA
GO_REF:0000117
ACCEPT
Summary: tRNA methylation, from an ARBA machine-learning model.
Reason: Correct core process, independently supported by IDA on this gene.
GO:0062152 mRNA (cytidine-5-)-methyltransferase activity
IEA
GO_REF:0000120
KEEP AS NON CORE
Summary: mRNA m5C methyltransferase activity from combined automatic methods.
Reason: GOA carries both sides of this question and both are retained. The catalytic assignment itself is supported by evidence that does not depend on the contested mapping assays: RNA-mediated activity-based protein profiling traps NSUN2 covalently on 5-fluorocytidine-containing RNA recovered by oligo-dT capture, and nucleoside LC-MS/MS of polyA-selected RNA shows an 82% fall in m5C in NSUN2 knockout cells (PMID:34556860); the mRNA export phenotype is rescued by wild-type but not methyltransferase-dead NSUN2 (PMID:28418038); and in vitro methylation of a defined p16 3'UTR fragment was reported (PMID:22395603). The activity is therefore not removed - and could not be, since this term also carries an EXP code whose full text has not been read here. What is disputed is the reliability and biological weight of the mRNA m5C maps. The methods disagree with one another (very low overlap between Aza-IP, bisulfite sequencing and miCLIP), mRNA secondary structure shields cytosines from bisulfite and generates false positives, and m5C occurs at roughly one in several thousand cytosines in mRNA, which leaves it unexplained how the proposed readers discriminate modified transcripts (PMID:42587746). The authors of the chemoproteomic study themselves note that polyA fractions carry ~40-fold lower m5C than total RNA and that contaminating tRNA/rRNA nucleosides could contribute to the measured signal. Demoted to non-core rather than accepted as core: the deeply conserved, structurally and genetically nailed-down substrate class for this Trm4 orthologue is tRNA (and vault RNA and mitochondrial tRNA), whereas the mRNA activity is quantitatively minor, its site maps are unresolved, and the large disease literature built on single mRNA m5C sites rarely supplies orthogonal validation.
GO:0005515 protein binding
IPI
PMID:33961781
Dual proteome-scale networks reveal cell-specific remodeling...
REMOVE
Summary: Bare protein binding from a proteome-scale affinity-purification interactome survey.
Reason: Per project curation guidance, protein binding carries no functional information and no partner is named by the term. Any biologically meaningful interaction of NSUN2 - for example with the m5C readers ALYREF and YBX1 - is better captured by the process terms for mRNA export and mRNA stability. No more informative molecular function is supportable from this evidence, so the row is removed rather than rewritten. Removal does not imply the reported interaction is false.
GO:0016428 tRNA (cytidine-N5)-methyltransferase activity
EXP
PMID:17071714
Identification of human tRNA:m5C methyltransferase catalysin...
ACCEPT
Summary: Experimental demonstration that the human Trm4 orthologue forms m5C34 in intron-containing pre-tRNA-Leu(CAA).
Reason: This is the founding evidence for the core molecular function, with substrate specificity defined at single-nucleotide resolution and cross-species complementation of a yeast trm4 deletion.
Supporting Evidence:
PMID:17071714
Using transcripts of intron-less and intron-containing human pre-tRNA Leu (CAA) genes as substrates, we have shown that m5C34 is introduced only in the intron-containing tRNA precursors when the substrates were incubated in the HeLa extract.
GO:0005737 cytoplasm
EXP
PMID:31276587
NSUN2 introduces 5-methylcytosines in mammalian mitochondria...
ACCEPT
Summary: Experimental localisation of NSUN2 including a cytoplasmic pool, in the study that traced its import into mitochondria.
Reason: Correct, if general.
GO:0062152 mRNA (cytidine-5-)-methyltransferase activity
EXP
PMID:34556860
Activity-based RNA-modifying enzyme probing reveals DUS3L-me...
KEEP AS NON CORE
Summary: Chemoproteomic profiling that traps m5C methyltransferases covalently on 5-fluorocytidine-containing RNA and recovers NSUN2 from an oligo-dT-selected fraction, with an 82% loss of polyA m5C on knockout.
Reason: GOA carries both sides of this question and both are retained. The catalytic assignment itself is supported by evidence that does not depend on the contested mapping assays: RNA-mediated activity-based protein profiling traps NSUN2 covalently on 5-fluorocytidine-containing RNA recovered by oligo-dT capture, and nucleoside LC-MS/MS of polyA-selected RNA shows an 82% fall in m5C in NSUN2 knockout cells (PMID:34556860); the mRNA export phenotype is rescued by wild-type but not methyltransferase-dead NSUN2 (PMID:28418038); and in vitro methylation of a defined p16 3'UTR fragment was reported (PMID:22395603). The activity is therefore not removed - and could not be, since this term also carries an EXP code whose full text has not been read here. What is disputed is the reliability and biological weight of the mRNA m5C maps. The methods disagree with one another (very low overlap between Aza-IP, bisulfite sequencing and miCLIP), mRNA secondary structure shields cytosines from bisulfite and generates false positives, and m5C occurs at roughly one in several thousand cytosines in mRNA, which leaves it unexplained how the proposed readers discriminate modified transcripts (PMID:42587746). The authors of the chemoproteomic study themselves note that polyA fractions carry ~40-fold lower m5C than total RNA and that contaminating tRNA/rRNA nucleosides could contribute to the measured signal. Demoted to non-core rather than accepted as core: the deeply conserved, structurally and genetically nailed-down substrate class for this Trm4 orthologue is tRNA (and vault RNA and mitochondrial tRNA), whereas the mRNA activity is quantitatively minor, its site maps are unresolved, and the large disease literature built on single mRNA m5C sites rarely supplies orthogonal validation.
Supporting Evidence:
PMID:34556860
Consistent with previous RNA bisulfite sequencing and m5C mass spectrometry that have implicated NSUN2 as the major mRNA m5C-forming enzyme16, 17, we found an 82% reduction in mRNA m5C levels upon NSUN2 knockout
PMID:34556860
We have employed a rigorous polyA mRNA purification, however since m5C is ~40-fold higher in total RNA than polyA-RNA, and comparably reduced in each sample upon NSUN2 KO (Figure 3b and 3c), we cannot exclude the possibility that contaminating tRNA/rRNA nucleosides may contribute to the measured m5C content.
GO:0005739 mitochondrion
HTP
PMID:34800366
Quantitative high-confidence human mitochondrial proteome an...
ACCEPT
Summary: Presence in a high-confidence quantitative human mitochondrial proteome.
Reason: Consistent with the two dedicated 2019 studies showing matrix import and mt-tRNA methylation; a high-throughput code here corroborates rather than substitutes for them.
GO:0006397 mRNA processing
IDA
PMID:22395603
The tRNA methyltransferase NSun2 stabilizes p16INK⁴ mRNA by ...
MODIFY
Summary: GO:0006397 mRNA processing covers conversion of a primary transcript into mature mRNA - splicing, capping, polyadenylation. The cited work reports mRNA stabilisation and nuclear export instead.
Reason: A term-choice problem rather than an evidence problem, so the annotations are modified rather than removed. PMID:22395603 reports that NSUN2 inhibits turnover of p16 mRNA, PMID:31358969 that NSUN2 with YBX1 stabilises HDGF mRNA, and PMID:28418038 that NSUN2-dependent m5C promotes ALYREF-mediated nuclear export. None of these is a pre-mRNA maturation step. The accurate terms are mRNA stabilization (GO:0048255) and regulation of mRNA export from nucleus (GO:0010793), the latter of which this gene already carries from PMID:28418038.
GO:0006397 mRNA processing
IDA
PMID:28418038
5-methylcytosine promotes mRNA export - NSUN2 as the methylt...
MODIFY
Summary: GO:0006397 mRNA processing covers conversion of a primary transcript into mature mRNA - splicing, capping, polyadenylation. The cited work reports mRNA stabilisation and nuclear export instead.
Reason: A term-choice problem rather than an evidence problem, so the annotations are modified rather than removed. PMID:22395603 reports that NSUN2 inhibits turnover of p16 mRNA, PMID:31358969 that NSUN2 with YBX1 stabilises HDGF mRNA, and PMID:28418038 that NSUN2-dependent m5C promotes ALYREF-mediated nuclear export. None of these is a pre-mRNA maturation step. The accurate terms are mRNA stabilization (GO:0048255) and regulation of mRNA export from nucleus (GO:0010793), the latter of which this gene already carries from PMID:28418038.
GO:0006397 mRNA processing
IDA
PMID:31358969
5-methylcytosine promotes pathogenesis of bladder cancer thr...
MODIFY
Summary: GO:0006397 mRNA processing covers conversion of a primary transcript into mature mRNA - splicing, capping, polyadenylation. The cited work reports mRNA stabilisation and nuclear export instead.
Reason: A term-choice problem rather than an evidence problem, so the annotations are modified rather than removed. PMID:22395603 reports that NSUN2 inhibits turnover of p16 mRNA, PMID:31358969 that NSUN2 with YBX1 stabilises HDGF mRNA, and PMID:28418038 that NSUN2-dependent m5C promotes ALYREF-mediated nuclear export. None of these is a pre-mRNA maturation step. The accurate terms are mRNA stabilization (GO:0048255) and regulation of mRNA export from nucleus (GO:0010793), the latter of which this gene already carries from PMID:28418038.
GO:0006400 tRNA modification
TAS
Reactome:R-HSA-6785438
ACCEPT
Summary: Reactome statement that NSUN2 methylates C40 and C48-C50 of tRNA-Gly(GCC).
Reason: Correct, and the specific child GO:0030488 tRNA methylation is also annotated to this gene.
GO:0016428 tRNA (cytidine-N5)-methyltransferase activity
IDA
PMID:34556860
Activity-based RNA-modifying enzyme probing reveals DUS3L-me...
ACCEPT
Summary: Mechanism-based covalent capture of NSUN2 on 5-fluorocytidine RNA, confirming the catalytic cysteine chemistry; NSUN2 knockout removes 76% of m5C from total RNA.
Reason: Core molecular function, supported by an assay that depends on catalysis.
Supporting Evidence:
PMID:34556860
We also found that NSUN2 installs the majority of m5C sites on total RNA (76% reduction upon NSUN2 KO) (Figure 3b), likely reflective of abundant tRNA m5C sites
GO:0005739 mitochondrion
IDA
PMID:31276587
NSUN2 introduces 5-methylcytosines in mammalian mitochondria...
ACCEPT
Summary: Proximity labelling and immunodetection place NSUN2 in the mitochondrial matrix.
Reason: Well-supported second location where NSUN2 methylates mt-tRNAs.
Supporting Evidence:
PMID:31276587
Here we employ spatially restricted proximity labelling and immunodetection to demonstrate that NSUN2 is imported into the matrix of mammalian mitochondria.
GO:0005739 mitochondrion
IDA
PMID:31287866
Mammalian NSUN2 introduces 5-methylcytidines into mitochondr...
ACCEPT
Summary: Structured illumination microscopy shows NSUN2 foci inside mitochondria.
Reason: Independent confirmation of the mitochondrial pool, from a different laboratory in the same year.
Supporting Evidence:
PMID:31287866
Although NSUN2 is predominantly localized to the nucleus and introduces m5C into cytoplasmic tRNAs and mRNAs, structured illumination microscopy clearly revealed NSUN2 foci inside mitochondria.
GO:0010793 regulation of mRNA export from nucleus
IDA
PMID:28418038
5-methylcytosine promotes mRNA export - NSUN2 as the methylt...
KEEP AS NON CORE
Summary: NSUN2-deposited m5C is read by ALYREF and promotes nuclear export of mRNA; the defect caused by NSUN2 depletion is rescued by wild-type but not catalytically dead NSUN2.
Reason: The catalytically dead rescue is the right control and makes this one of the better-supported mRNA m5C phenotypes, so the annotation is kept. It is placed as non-core for the same reason as the mRNA methyltransferase activity itself: it is downstream of a quantitatively minor substrate class whose site maps remain contested, and a reader operating at roughly one m5C per several thousand cytosines is mechanistically unexplained (PMID:42587746). The core of this gene is tRNA modification.
Supporting Evidence:
PMID:28418038
Dysregulation of ALYREF-mediated mRNA export upon NSUN2 depletion could be restored by reconstitution of wild-type but not methyltransferase-defective NSUN2.
GO:0016428 tRNA (cytidine-N5)-methyltransferase activity
IDA
PMID:31287866
Mammalian NSUN2 introduces 5-methylcytidines into mitochondr...
ACCEPT
Summary: m5C at positions 48-50 of mitochondrial tRNAs reconstituted in vitro with purified NSUN2 and SAM, and absent from Nsun2 knockout mouse and human cells.
Reason: The strongest single piece of evidence for the core molecular function: a defined enzyme, a defined substrate, mapped product positions, and loss of the product on gene knockout.
Supporting Evidence:
PMID:31287866
In addition, we successfully reconstituted m5C at positions 48-50 of mt-tRNA in vitro with NSUN2 protein in the presence of S-adenosylmethionine.
GO:0030488 tRNA methylation
IDA
PMID:31287866
Mammalian NSUN2 introduces 5-methylcytidines into mitochondr...
ACCEPT
Summary: NSUN2-dependent m5C mapped by mass spectrometry in eight mouse and six human mt-tRNAs.
Reason: Core biological process.
Supporting Evidence:
PMID:31287866
In this study, we took advantage of mass spectrometric analysis to map 5-methylcytidine (m5C) at positions 48-50 in eight mouse and six human mt-tRNAs.
GO:0036416 tRNA stabilization
ISS
GO_REF:0000024
ACCEPT
Summary: m5C deposited by NSUN2 protects tRNAs from stress-induced endonucleolytic cleavage.
Reason: A direct consequence of the core catalytic activity rather than a separate role, and consistent with the loss of C47/C48 methylation on tRNA-Asp in NSUN2-null patient cells.
GO:0048820 hair follicle maturation
ISS
GO_REF:0000024
KEEP AS NON CORE
Summary: Hair follicle maturation, transferred from the mouse Nsun2 knockout skin phenotype.
Reason: A real organismal phenotype, but a distal consequence of tRNA hypomodification in a particular stem-cell compartment rather than an activity of the protein.
GO:0062152 mRNA (cytidine-5-)-methyltransferase activity
IDA
PMID:22395603
The tRNA methyltransferase NSun2 stabilizes p16INK⁴ mRNA by ...
KEEP AS NON CORE
Summary: In vitro methylation of the p16 3'UTR by NSun2, with knockdown shortening and overexpression lengthening the p16 mRNA half-life.
Reason: GOA carries both sides of this question and both are retained. The catalytic assignment itself is supported by evidence that does not depend on the contested mapping assays: RNA-mediated activity-based protein profiling traps NSUN2 covalently on 5-fluorocytidine-containing RNA recovered by oligo-dT capture, and nucleoside LC-MS/MS of polyA-selected RNA shows an 82% fall in m5C in NSUN2 knockout cells (PMID:34556860); the mRNA export phenotype is rescued by wild-type but not methyltransferase-dead NSUN2 (PMID:28418038); and in vitro methylation of a defined p16 3'UTR fragment was reported (PMID:22395603). The activity is therefore not removed - and could not be, since this term also carries an EXP code whose full text has not been read here. What is disputed is the reliability and biological weight of the mRNA m5C maps. The methods disagree with one another (very low overlap between Aza-IP, bisulfite sequencing and miCLIP), mRNA secondary structure shields cytosines from bisulfite and generates false positives, and m5C occurs at roughly one in several thousand cytosines in mRNA, which leaves it unexplained how the proposed readers discriminate modified transcripts (PMID:42587746). The authors of the chemoproteomic study themselves note that polyA fractions carry ~40-fold lower m5C than total RNA and that contaminating tRNA/rRNA nucleosides could contribute to the measured signal. Demoted to non-core rather than accepted as core: the deeply conserved, structurally and genetically nailed-down substrate class for this Trm4 orthologue is tRNA (and vault RNA and mitochondrial tRNA), whereas the mRNA activity is quantitatively minor, its site maps are unresolved, and the large disease literature built on single mRNA m5C sites rarely supplies orthogonal validation.
Supporting Evidence:
PMID:22395603
In vitro methylation assays show that NSun2 methylates the p16 3'UTR at A988.
GO:0062152 mRNA (cytidine-5-)-methyltransferase activity
IDA
PMID:28418038
5-methylcytosine promotes mRNA export - NSUN2 as the methylt...
KEEP AS NON CORE
Summary: NSUN2 identified as the main enzyme forming m5C in mRNA, with transcriptome-wide maps and a catalytically dead rescue control.
Reason: GOA carries both sides of this question and both are retained. The catalytic assignment itself is supported by evidence that does not depend on the contested mapping assays: RNA-mediated activity-based protein profiling traps NSUN2 covalently on 5-fluorocytidine-containing RNA recovered by oligo-dT capture, and nucleoside LC-MS/MS of polyA-selected RNA shows an 82% fall in m5C in NSUN2 knockout cells (PMID:34556860); the mRNA export phenotype is rescued by wild-type but not methyltransferase-dead NSUN2 (PMID:28418038); and in vitro methylation of a defined p16 3'UTR fragment was reported (PMID:22395603). The activity is therefore not removed - and could not be, since this term also carries an EXP code whose full text has not been read here. What is disputed is the reliability and biological weight of the mRNA m5C maps. The methods disagree with one another (very low overlap between Aza-IP, bisulfite sequencing and miCLIP), mRNA secondary structure shields cytosines from bisulfite and generates false positives, and m5C occurs at roughly one in several thousand cytosines in mRNA, which leaves it unexplained how the proposed readers discriminate modified transcripts (PMID:42587746). The authors of the chemoproteomic study themselves note that polyA fractions carry ~40-fold lower m5C than total RNA and that contaminating tRNA/rRNA nucleosides could contribute to the measured signal. Demoted to non-core rather than accepted as core: the deeply conserved, structurally and genetically nailed-down substrate class for this Trm4 orthologue is tRNA (and vault RNA and mitochondrial tRNA), whereas the mRNA activity is quantitatively minor, its site maps are unresolved, and the large disease literature built on single mRNA m5C sites rarely supplies orthogonal validation.
Supporting Evidence:
PMID:28418038
Moreover, m5C formation in mRNAs is mainly catalyzed by the RNA methyltransferase NSUN2, and m5C is specifically recognized by the mRNA export adaptor ALYREF as shown by in vitro and in vivo studies.
GO:0062152 mRNA (cytidine-5-)-methyltransferase activity
IDA
PMID:31358969
5-methylcytosine promotes pathogenesis of bladder cancer thr...
KEEP AS NON CORE
Summary: Single-nucleotide-resolution mRNA m5C landscape in bladder cancer, with NSUN2 as writer and YBX1 as reader on the HDGF 3'UTR.
Reason: GOA carries both sides of this question and both are retained. The catalytic assignment itself is supported by evidence that does not depend on the contested mapping assays: RNA-mediated activity-based protein profiling traps NSUN2 covalently on 5-fluorocytidine-containing RNA recovered by oligo-dT capture, and nucleoside LC-MS/MS of polyA-selected RNA shows an 82% fall in m5C in NSUN2 knockout cells (PMID:34556860); the mRNA export phenotype is rescued by wild-type but not methyltransferase-dead NSUN2 (PMID:28418038); and in vitro methylation of a defined p16 3'UTR fragment was reported (PMID:22395603). The activity is therefore not removed - and could not be, since this term also carries an EXP code whose full text has not been read here. What is disputed is the reliability and biological weight of the mRNA m5C maps. The methods disagree with one another (very low overlap between Aza-IP, bisulfite sequencing and miCLIP), mRNA secondary structure shields cytosines from bisulfite and generates false positives, and m5C occurs at roughly one in several thousand cytosines in mRNA, which leaves it unexplained how the proposed readers discriminate modified transcripts (PMID:42587746). The authors of the chemoproteomic study themselves note that polyA fractions carry ~40-fold lower m5C than total RNA and that contaminating tRNA/rRNA nucleosides could contribute to the measured signal. Demoted to non-core rather than accepted as core: the deeply conserved, structurally and genetically nailed-down substrate class for this Trm4 orthologue is tRNA (and vault RNA and mitochondrial tRNA), whereas the mRNA activity is quantitatively minor, its site maps are unresolved, and the large disease literature built on single mRNA m5C sites rarely supplies orthogonal validation.
Supporting Evidence:
PMID:31358969
Moreover, NSUN2 and YBX1 are demonstrated to drive UCB pathogenesis by targeting the m5C methylation site in the HDGF 3' untranslated region.
GO:0070062 extracellular exosome
IDA
PMID:28341602
Cytosolic YB-1 and NSUN2 are the only proteins recognizing s...
KEEP AS NON CORE
Summary: NSUN2 detected in exosomes secreted by HEK293 cells, where it binds a specific RNA motif.
Reason: The observation is real but peripheral; it says where some of the protein ends up, not what the enzyme does. The accompanying motif-binding result is better expressed as RNA binding, which this gene already carries.
Supporting Evidence:
PMID:28341602
Both these proteins are found in exosomes secreted by HEK293 cells.
GO:2000736 regulation of stem cell differentiation
ISS
GO_REF:0000024
KEEP AS NON CORE
Summary: Regulation of stem cell differentiation, transferred from mouse.
Reason: Genuine but downstream: NSUN2 loss perturbs differentiation through altered tRNA methylation, translation and stress-induced tRNA fragments, not through a dedicated regulatory activity.
GO:0005654 nucleoplasm
TAS
Reactome:R-HSA-6785438
ACCEPT
Summary: Nucleoplasmic location asserted by the Reactome tRNA modification reaction.
Reason: Consistent with the nuclear localisation of the enzyme and of pre-tRNA methylation.
GO:0003723 RNA binding
HDA
PMID:22658674
Insights into RNA biology from an atlas of mammalian mRNA-bi...
ACCEPT
Summary: Recovery of NSUN2 in mRNA interactome capture experiments.
Reason: Unlike bare protein binding, RNA binding is informative here: it is the substrate-recognition function of an RNA-modifying enzyme, and NSUN2 also binds RNA in contexts where it is not obviously catalysing, such as the exosomal motif recognition reported in PMID:28341602.
GO:0003723 RNA binding
HDA
PMID:22681889
The mRNA-bound proteome and its global occupancy profile on ...
ACCEPT
Summary: Recovery of NSUN2 in mRNA interactome capture experiments.
Reason: Unlike bare protein binding, RNA binding is informative here: it is the substrate-recognition function of an RNA-modifying enzyme, and NSUN2 also binds RNA in contexts where it is not obviously catalysing, such as the exosomal motif recognition reported in PMID:28341602.
GO:0005730 nucleolus
IDA
PMID:17071714
Identification of human tRNA:m5C methyltransferase catalysin...
ACCEPT
Summary: Immunolocalisation of the human Trm4 orthologue to the nucleolus.
Reason: Correct and specific primary location.
GO:0005737 cytoplasm
IDA
PMID:17071714
Identification of human tRNA:m5C methyltransferase catalysin...
ACCEPT
Summary: A cytoplasmic pool of the enzyme was also seen.
Reason: Correct, if general.
GO:0016428 tRNA (cytidine-N5)-methyltransferase activity
IDA
PMID:17071714
Identification of human tRNA:m5C methyltransferase catalysin...
ACCEPT
Summary: Direct assay of m5C34 formation on human pre-tRNA-Leu(CAA).
Reason: Core molecular function.
GO:0030488 tRNA methylation
IDA
PMID:17071714
Identification of human tRNA:m5C methyltransferase catalysin...
ACCEPT
Summary: tRNA methylation demonstrated directly for the human enzyme.
Reason: Core biological process.

Core Functions

NSUN2 is the human Trm4 orthologue and the principal RNA m5C writer of the cell. Using SAM and a motif VI catalytic cysteine, it methylates C34 of intron-containing pre-tRNA-Leu(CAA) and C47-C50 in the variable loop of many cytoplasmic tRNAs, and the same enzyme, imported into the mitochondrial matrix, makes m5C at positions 48-50 of several mitochondrial tRNAs. The activity has been reconstituted in vitro with purified protein and mapped product positions, and the corresponding modifications are lost in NSUN2-null mouse tissue, NSUN2 knockout human cells and cells from patients with biallelic NSUN2 variants. The methylation stabilises tRNA against stress-induced endonucleolytic cleavage.

Supporting Evidence:
  • PMID:17071714
    Using transcripts of intron-less and intron-containing human pre-tRNA Leu (CAA) genes as substrates, we have shown that m5C34 is introduced only in the intron-containing tRNA precursors when the substrates were incubated in the HeLa extract.
  • PMID:31287866
    In addition, we successfully reconstituted m5C at positions 48-50 of mt-tRNA in vitro with NSUN2 protein in the presence of S-adenosylmethionine.
  • PMID:22577224
    Patient cells lacked NSUN2 protein and there was resultant loss of site-specific 5-cytosine methylation of the tRNA(Asp GTC) at C47 and C48, known NSUN2 targets.

Substrate recognition by NSUN2 extends beyond tRNA. It is the only enzyme that methylates C69 of the vault RNA VTRNA1.1, and that single modification determines how VTRNA1.1 is processed into small vault RNAs, so NSUN2 binding and methylation of a structured non-coding RNA has a defined downstream consequence. NSUN2 is also recovered in mRNA interactome capture and recognises a specific structural motif in exosome-enriched transcripts. This binding function is listed separately from the catalytic function because in several of these contexts the binding, rather than the chemistry, is what has been directly demonstrated.

Molecular Function:
RNA binding
Cellular Locations:
Supporting Evidence:
  • PMID:31186410
    Methylation of cytosine 69 in VTRNA1.1 occurs frequently in human cells, is exclusively mediated by NSUN2, and determines the processing of VTRNA1.1 into small-vault RNAs (svRNAs).
  • PMID:28341602
    Both these proteins are found in exosomes secreted by HEK293 cells.

References

Loading supporting content…

Download this section (compressed HTML)

Suggested Questions for Experts

Q: Is NSUN2-dependent m5C in mRNA a physiologically significant modification in its own right, or largely the low-level off-target activity of an enzyme whose real substrate is tRNA? The nucleoside mass-spectrometry evidence for NSUN2-dependent m5C in polyA RNA is real, but polyA fractions carry about forty-fold less m5C than total RNA and the authors of that study cannot exclude contaminating tRNA and rRNA nucleosides. What level of polyA purity, or what orthogonal single-molecule method, would settle this?

Suggested experts: Ralph E Kleiner, Michaela Frye

Q: Which of the thousands of reported mRNA m5C sites survive a method-agnostic test? Aza-IP, bisulfite sequencing and miCLIP overlap very poorly, and mRNA secondary structure produces false-positive bisulfite signals. Is there a consensus high-confidence site set for NSUN2, and what is its size?

Suggested experts: Alexey A Malygin

Q: If m5C occurs at roughly one in several thousand cytosines in mRNA, and ALYREF and YBX1 bind m5C-containing regions only a few fold better than unmodified ones, how can a reader-based mechanism achieve specificity in vivo? Does the reader model require additional co-recruitment, or should the phenotypes attributed to single mRNA m5C sites be re-examined?

Suggested experts: Alexey A Malygin, Yun-Gui Yang

Q: Does NSUN2 have any role in ribosomal RNA or in mitochondrial ribosome assembly, as its IBA annotations assert? These were propagated from mouse Nsun4, a paralogue with a distinct substrate and complex, and NSUN2 loss does not appreciably impair mitochondrial translation. Should the PAINT node for these terms be moved?

Q: How much of the NSUN2 loss-of-function phenotype in human intellectual disability is attributable to tRNA hypomodification and stress-induced tRNA fragments, and how much, if any, to loss of mRNA or vault RNA methylation? A separation-of-function allele would be the cleanest test.

Suggested experts: Michaela Frye, Joseph G Gleeson

Suggested Experiments

Experiment: Apply direct RNA nanopore sequencing, or another method that does not require chemical conversion or antibody enrichment, to matched wild-type and NSUN2 knockout human cells, with a catalytically dead NSUN2 add-back as a third arm. Intersect the resulting site calls with published bisulfite, Aza-IP and miCLIP sets and report the size of the reproducible core. Include synthetic spike-ins of known modification stoichiometry to calibrate false-positive and false-negative rates.

Hypothesis: NSUN2-dependent m5C sites in mRNA are far fewer than current maps report, and only a small high-confidence subset is reproducible across methods.

Type: Direct RNA sequencing

Experiment: Quantify m5C by LC-MS/MS in polyA RNA purified to increasing stringency (successive oligo-dT rounds, then size selection, then depletion with tRNA- and rRNA-specific probes), measuring residual tRNA and rRNA in each fraction by qPCR, and plot m5C against measured contamination to extrapolate the true mRNA value.

Hypothesis: The mRNA m5C signal measured by nucleoside mass spectrometry in polyA fractions is partly contributed by contaminating tRNA and rRNA.

Type: Quantitative mass spectrometry

Experiment: For a small number of prominent claimed sites (for example HDGF and IL1B), introduce a silent or minimally disruptive C-to-U or C-to-A substitution at the endogenous locus by base editing, and test whether transcript stability and the downstream phenotype change as predicted, with NSUN2 knockout as a positive control. This replaces enzyme-level perturbation, which affects thousands of transcripts, with site-level perturbation.

Hypothesis: Individual mRNA m5C sites proposed to drive disease phenotypes are functional at that site.

Type: Base editing and phenotyping

Experiment: Screen NSUN2 variants, guided by the tRNA-bound structure, for alleles that retain tRNA C34 and C48-C50 methylation while losing mRNA methylation (or the reverse), validating each by nucleoside LC-MS/MS on separated RNA fractions. Introduce any separation-of-function allele into cells and mice and ask which arm accounts for the differentiation and neurodevelopmental phenotypes.

Hypothesis: The tRNA and mRNA functions of NSUN2 can be genetically separated.

Type: Structure-guided mutagenesis and mouse genetics

Experiment: Test directly for NSUN2-dependent rRNA m5C: map m5C in cytoplasmic and mitochondrial rRNA by mass spectrometry in wild-type and NSUN2 knockout cells, and assay mitoribosome assembly by sucrose-gradient profiling in the same cells, using NSUN4 knockout as the positive control.

Hypothesis: The rRNA-related IBA annotations on NSUN2 are paralogue leakage from NSUN4.

Type: Mass spectrometry and ribosome profiling

📚 Additional Documentation

Notes

(NSUN2-notes.md)

Loading supporting content…

Download this section (compressed HTML)

📄 View Raw YAML

Loading supporting content…

Download this section (compressed HTML)