Article Text

Download PDFPDF

Short report
TRMT10A dysfunction is associated with abnormalities in glucose homeostasis, short stature and microcephaly
  1. David Gillis1,
  2. Aiswarya Krishnamohan2,
  3. Barak Yaacov3,
  4. Avraham Shaag3,
  5. Jane E Jackman2,
  6. Orly Elpeleg3
  1. 1Department of Pediatrics, Hadassah-Hebrew University Medical Center, Ein-Kerem, Jerusalem, Israel
  2. 2Department of Chemistry and Biochemistry, Ohio State Biochemistry Program and Center for RNA Biology, The Ohio State University, Columbus, Ohio, USA
  3. 3Monique and Jacques Roboh Department of Genetic Research, Hadassah-Hebrew University Medical Center, Jerusalem, Israel
  1. Correspondence to Professor Orly Elpeleg, Monique and Jacques Roboh Department of Genetic Research, Hadassah-Hebrew University Medical Center, Jerusalem 91120, Israel; Elpeleg{at}


Background Trm10 is a tRNA m1G9 methyltransferase, which in yeast modifies 12 different tRNA species, yet is considered non-essential for viability under standard growth conditions. In humans, there are three Trm10 orthologs, one mitochondrial and two presumed cytoplasmic. A nonsense mutation in one of the cytoplasmic orthologs (TRMT10A) has recently been associated with microcephaly, intellectual disability, short stature and adolescent onset diabetes.

Methods and results The subjects were three patients who suffered from microcephaly, intellectual disability, short stature, delayed puberty, seizures and disturbed glucose metabolism, mainly hyperinsulinaemic hypoglycaemia. A homozygous Gly206Arg (G206R) mutation in the TRMT10A gene was identified using whole exome sequencing. The mutation segregated in the family and was absent from large control cohorts. Determination of the methylation activity of the expressed wild-type (WT) and variant TRMT10A enzymes with transcripts of 32P -tRNAGlyGCC as a substrate revealed a striking defect (<0.1% of WT activity) for the variant enzyme. The binding affinity of the G206R variant enzyme to tRNA, determined by fluorescence anisotropy, was similar to that of the WT enzyme.

Conclusions The completely abolished m1G9 methyltransferase activity of the mutant enzyme is likely due to significant defects in its ability to bind the methyl donor S-adenosyl methionine. We propose that TRMT10A deficiency accounts for abnormalities in glucose homeostasis initially manifesting both ketotic and non-ketotic hypoglycaemic events with transition to diabetes in adolescence, perhaps as a consequence of accelerated β cell apoptosis. The seizure disorder and intellectual disability are probably secondary to mutant gene expression in neuronal tissue.

  • Endocrinology

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.


Post-transcriptional modifications of RNA are ubiquitous in biology, and among RNA species, tRNA are by far the most extensively modified of the cellular RNAs. A typical eukaryotic tRNA contains more than 10 modified nucleotides of differing types and more than 100 distinct chemical modifications of tRNA nucleotide bases and sugars have been identified to date.1 ,2 A correspondingly large number of enzymes have been identified that catalyse the numerous post-transcriptional tRNA modification reactions and these enzymes are often highly conserved, catalysing similar modifications in multiple domains of life.2–4

Trm10 is a tRNA m1G9 methyltransferase first identified in yeast, but its homologues are widespread throughout eukarya and archaea.5 ,6 In humans, there are three Trm10 orthologs with one (TRMT10A) most closely related to yeast Trm10 in overall sequence.5 ,7 ,8 Trm10 in yeast catalyses modification of at least 12 different tRNA species, and yet deletion of the TRM10 gene does not cause any observable growth defect of the corresponding strains under standard conditions.5–9

We now report the identification of a homozygous missense mutation in the TRMT10A gene in three patients who suffered from intellectual disability, short stature and disturbed glucose metabolism. Here we show that introduction of the mutation p.G206R into human TRMT10A completely abolishes m1G9 methyltransferase activity of the purified protein in vitro, which may be due to significant defects in the ability of the mutant enzyme to bind the methyl donor S-adenosyl methionine (SAM).

Subjects and methods


The three were a female and two male patients, the third, eighth and ninth of 12 siblings born to non-consanguineous parents, who both originated from the small, isolated, inbred Jewish community of Uzbekistan. All were born at an unknown gestational age; the birth weights were reported as low (1.7–2.4 kg) and relative microcephaly (31.3–31.5 cm) was evident at birth. All grew along low growth centiles with head circumference relatively lower (−2.5 to −3.94 SD) than their weight and height during infancy (figure 1). Psychomotor development was delayed and the patients always attended the intellectual disabilities schooling system and were toilet trained only at 6–7 years of age. The patients were intermittently lost to follow-up but at the time of reporting, aged 19, 14 and 13, they still required assistance while dressing, and would run away with strangers if left unattended. Recurrent seizures were documented in all three children from 5 to 6 years of age; EEG recordings of all three patients were normal. Brain MRI performed in patient II-3 at 6 years of age and a computerised tomographic scan of patient II-8 at 5 years of age with and without contrast were also normal. In addition to the short stature, patient II-3, a female patient, had delayed pubertal development at 16 years of age, with breast Tanner stage 2 and pubic hair stage 2; there were still no periods at 19 years of age. The parents and the other children in the family were healthy; specifically, the father’s height was 163 cm and the mother's height was 158 cm and neither parent nor any of the healthy siblings was diabetic. Metabolic studies that allowed some insight into the insulin secretory pattern of these patients are described in the Results section.

Figure 1

Growth centiles of the three patients.


Whole exome analysis

Exonic sequences were enriched in the DNA sample of patient II-3 using SureSelect Human All Exon 50 Mb Kit (AgilentTechnologies, Santa Clara, California, USA). Sequences were determined by HiSeq2000 (Illumina, San Diego, California, USA) and 100 bp were read paired-end. Reads alignment and variant calling were performed with DNAnexus software (Palo Alto, California, USA) using the default parameters with the human genome assembly hg19 (GRCh37) as a reference. Patient and parental consent was given for DNA studies. The study was performed with the approval of the ethical committees of Hadassah Medical Center and the Ministry of Health.

In vitro methylation activity assay

The G206R variant was introduced by Phusion mutagenesis (Thermo Scientific) into a plasmid previously used for purification of human TRMT10A and clones were verified by sequencing. Wild-type and G206R variant TRMT10A enzymes were expressed and purified from Escherichia coli as previously described.9 Methylation activity was tested with in vitro transcripts of tRNAGlyGCC, which were specifically labelled with 32P at the phosphate immediately 5′-to G9 by published methods.9 The labelled tRNAGly was then tested in methylation assays using 10-fold serial dilutions (3 µM–3 nM) of either wild-type or G206R human TRMT10A enzymes. The reaction was conducted at 30°C in a reaction buffer containing 50 mM Tris pH 8, 1.5 mM MgCl2 and 0.5 mM SAM. After 2 h, the reaction was stopped by the addition of phenol:chloroform:isoamyl alcohol (25:24:1) and reaction products were purified by phenol extraction followed by ethanol precipitation. The precipitated RNA-containing pellet was resuspended in Nuclease P1 digestion buffer containing 0.25 mg/mL Nuclease P1 (Sigma-Aldrich), and then digested for 1 h. Nuclease P1 cleaves the labelled RNA into individual 5′-monophosphorylated nucleotides, and thus releases either labelled p*G9 derived from unreacted substrate or labelled p*m1G9 derived from the methylated tRNA. These two products are resolved by cellulose thin-layer chromatography in a solvent containing 66% isobutyric acid, 33% H2O and 1% NH4OH.

Fluorescence Anisotropy to test binding affinity of TRMT10A for tRNAGly

The binding of wild-type and G206R TRMT10A to tRNAGly was tested with a tRNAGly transcript that is 5′-labelled with 6-carboxyfluorescein (6-FAM). The fluorescently-labelled tRNA was produced by ligation of a 5′-FAM-labelled 8-mer oligonucleotide corresponding to the first eight nucleotides of tRNAGly with an in vitro transcript corresponding to the remaining nucleotides of the tRNA, as described previously for production of radiolabelled tRNA.5 ,9 The fluorescently-labelled tRNAGly (15 nM) was incubated with varying concentrations (50–1000 nM) of purified enzyme as indicated; bovine serum albumin was used as a control protein to detect non-specific binding. The binding buffer contained 50 mM bis-tris pH 6.0 and 1.5 mM MgCl2. Binding reactions were carried out for 30 min at room temperature protected from light, then transferred into a 384-well black polystyrene microplate (Corning) for measurement of fluorescence anisotropy (FA) using the Infinite M1000 PRO fluorometer (Tecan). Excitation was at 470 nm and emission intensity (at 518 nm) was measured from the parallel and perpendicular planes. Apparent equilibrium dissociation constants (KD) were determined by fitting the observed FA (from two independent experiments) as a function of protein concentration (C) to a binding equation for cooperative binding (equation 1), where FAmin and FAmax are the minimum and maximum limits to the observed FA and n is the Hill coefficient, using Kaleidagraph (Synergy software).Embedded Image 1


Studies of glucose and insulin in the siblings

There was no available evidence of hypoglycaemia for any of the three siblings in the neonatal period. Documentation of hypoglycaemia and studies thereof are presented here for each of the three patients separately.

Patient II-3: Relatively low postseizure blood glucose levels of about 70 mg/dL were documented after seizures in this patient from 5 years of age. However, since preseizure levels were, of course, not available, the presence of preseizure hypoglycaemia of below 50 mg/dL was not documented until close monitoring of blood glucose levels during a prolonged hospital admission at 9 years of age. During this admission, postprandial hyperglycaemia (>250 mg/dL), followed within 3–4 h by severe hypoglycaemia (<40 mg/dL), was documented. Ketotic hypoglycaemia was evident on one occasion after a 3 h fast (glucose 1.7 mmol/L (30.6 mg/dL) with undetectable insulin and c-peptide, acetoacetate 780 µmol/L, β-hydroxybutyrate 3356 µmole/L, free fatty acids (FFA) 919 µmole/L). During the same episode, cortisol was 524 mmol/L, and growth hormone 8.5 ng/mL thus ruling out both cortisol and growth hormone deficiency. A serum cortisol level of 1073 nmol/L 1 h after intravenous aqueous adrenocorticotropin (0.25 mg/m2 body surface area) was then determined to further assure absence of adrenal insufficiency.

A formal oral glucose tolerance test (OGTT) during the same admission showed fasting insulin above normal, despite low-normal fasting blood glucose. Although there were technical difficulties with the test that did not allow the usual 30- and 60-min samples to be taken, the glucose levels were high and did not start to decline until the 110-min sample. By 180 min, glucose levels were still high with insulin levels that were not unusually elevated and obviously not high enough to normalise serum glucose (figure 2). Taken together, the OGTT findings were consistent with insulin resistance (high fasting insulin but normal fasting glucose levels) along with inappropriately low insulin secretion in response to glucose. GLUT2 and GYS1 defects were excluded by normal oral galactose tolerance test and absence of mutations in these genes upon whole exome analysis. The patient was treated with high fat and protein diet and with very low carbohydrate intake and this, together with nocturnal formula feeding, prevented hypoglycaemia. Occasional seizures were still reported following high carbohydrate meals taken against medical advice. At 19 years of age, 10 years later, upon testing at the hostel where she resides, normal fasting blood glucose levels were noted and high 2-h postprandial glucose levels were reported (248 mg/dL and 176 mg/dL on two separate days).

Figure 2

Formal oral glucose tolerance test in patient II-3 at 9 years of age. Note low-normal fasting blood glucose (3.9 mmol/L) with non-suppression of insulin (204 pmol/L) and glucose of 9.8 mmol/L at 180 min indicating inadequate glucose handling.

Patient II-8 : On three separate occasions, when tested during hospital admission because of observed seizure activity at 6 years of age, insulin levels were 720, 923 and 659 pmol/L, during hypoglycaemia <36 mg/dL on all three occasions (normal fasting insulin≤110 pmol/L) accompanied by high c-peptide level and low serum ketones. These hypoglycaemic episodes were not associated with formal fasting and there was no clear documentation of the time lapse from the last meal. This patient was treated for some time successfully with oral diazoxide.

Patient II-9: Hypoglycaemia was first documented at 5 years of age. A formal OGTT was performed at 6 years of age. The fasting (time 0) plasma glucose result of 1.8 mmol/L was clearly in the hypoglycaemic range and the test was terminated at 150 min because of hypoglycaemia (2.2 mmol/L); at termination of the test, insulin was still measurable indicating that he also had inappropriate insulin secretion. The parents refused further investigations and the patient was treated by frequent feeds only.

While this manuscript was in preparation, a 13th child, a boy, was born to the family and was found to be homozygous for the G206R mutation in the TRMT10A gene (see below). Fasting blood glucose was tested on several occasions, last at 10 months of age and was repeatedly normal. At 9 months, he was microcephalic (head circumference 42 cm, −2.32 SDs from average for boys) and had mild developmental delay.

Genetic studies

In order to identify the disease-causing gene in this family, we performed whole exome sequencing in the DNA sample of patient II-3. Aligning the patient 43.0 million reads to the reference human genome revealed 134,241 variants. We removed variants which were of low depth (<x8), deep intronic, and those present in dbSNP132 or in the inhouse dbSNP. Because of the common ethnic origin of the parents, we focused on homozygous variants which were predicted pathogenic by Mutation Taster software.10 Only two homozygous variants survived this filtering process, chr9:35704037 A>T, p. Val2061Glu in TLN1 gene and chr4: 100474971 C>T, (c.616G>A), p.Gly206Arg (G206R) in the RG9MTD2 (TRMT10A) gene. Genotyping the 14 family members excluded the TLN1 mutation and confirmed full segregation of the TRMT10A mutation with the disease in this family (figure 3A–D). The mutation was absent from dbSNP138 and from the 6503 healthy individuals whose exome analysis results were available through the Exome Variant Server, NHLBI Exome Sequencing Project, Seattle, Washington, USA ( EVS-V.0.0.21) (accessed 30 November 2013). We were unfortunately unable to identify sufficient samples from healthy people originating from the same Jewish community.

Figure 3

(A) Pedigree of the family and the genotype of the TRMT10A mutation. Filled symbols denote affected individuals. (B–D) Sequence of part of exon 6 of the TRMT10A gene. The mutation site is indicated (arrow) in the patient (B), mother (C) and healthy sister II-10. (E) The evolutionary conservation of Gly206 (in red) in the TRMT10A protein.

TRMT10A contains eight exons, the first exon being non-protein coding. It encodes 340 amino acids and the glycine at position 206 is conserved throughout evolution (figure 3E). In order to study the functional significance of the mutation, we reconstructed the G206R mutation in the context of human TRMT10A and purified the variant protein for testing with in vitro activity assays. The G206R variant purified with similar yield to that of the wild-type enzyme, suggesting that the overall structure of the enzyme was not substantially altered upon introduction of the mutation. Moreover, in vitro tRNA binding assays showed nearly identical ability of the wild-type and G206R variant enzyme to bind to a known tRNA substrate for m1G9 methylation (yeast tRNAGlyGCC), as judged by the similar KD,app exhibited by both enzymes (figure 4A). Therefore, although levels of TRMT10A protein in the patient cells have not been measured directly, these data suggest that the G206R mutation is unlikely to significantly affect TRMT10A protein levels as a consequence of protein instability caused by folding defects associated with the mutation. However, upon measuring the ability of the G206R TRMT10A variant to methylate the same tRNA using a previously described in vitro activity assay, a striking reduction (at least 104-fold) in methylation activity was observed for the variant enzyme (figure 4B).

Figure 4

Biochemical characterisation of the G206R TRMT10A variant. (A) Binding of TRMT10A protein to fluorescently-labelled tRNAGlyGCC measured using fluorescence anisotropy (FA). Binding assays were performed with limiting tRNA (15 nM) and wild-type TRMT10A (filled squares); G206R TRMT10A (open squares) and bovine serum albumin (BSA) (filled circles) as a control. FA was measured according to methods, and the KD,app was derived from fits to the data; wild-type KD,app=220±20 nM; G206R KD,app=190±12 nM. (B) In vitro methylation activity of TRMT10A with tRNAGlyGCC substrate. The tRNA (uniquely labelled at G9) was incubated with 10-fold serial dilutions of enzyme (3 µM–3 nM final concentrations) in the presence of S-adenosyl methionine. After digestion of the reaction products with nuclease P1, labelled p*G9 (derived from unreacted substrate) is resolved from labelled p*m1G9 (the methylated reaction product) by thin-layer chromatography. NE, no enzyme control lane.

Since methylation was completely undetectable even at the highest concentrations of the mutant enzyme in the assays (figure 4B), it was impossible to further characterise the specific catalytic defect associated with the mutation (although the proficient tRNA binding affinity demonstrated above appears to rule out a loss in tRNA affinity as the source of the inactivity). However, a hypomorphic variant of yeast Trm10 with alterations of the analogous G206 residue and the succeeding G207 residue, both to alanine, exhibits significantly compromised ability to bind to the SAM methyl donor compared with the wild-type enzyme (data not shown), and the location of the G206 residue in the SAM-binding pocket observed in the published Trm10 crystal structure11 suggests that the loss of catalytic activity with the G206R mutation may be specifically due to inability of the variant enzyme to bind the methyl donor.


While this manuscript was in preparation, three siblings with a rather similar phenotype were reported to have a nonsense mutation in the TRMT10A gene.12 The missense mutation in TRMT10A identified here appears to phenocopy the complete loss of TRMT10A (since the nonsense mutation would predictably produce a truncated protein containing less than 30% of the wild-type enzyme), effectively ruling out the possibility that the disease phenotype is associated with indirect effects due to loss of interactions between TRMT10A and other cellular components. Instead, these data strongly support the fact that loss of methyltransferase activity and a consequent decrease in the amount of m1G9-methylated tRNA are responsible for the observed pathogenicity.

Our patients shared the neuronal and growth failure described in association with the nonsense mutation in TRM10A, but their glucose metabolism disturbances, including the ketotic, non-ketotic and hyperinsulinaemic hypoglycaemia and postprandial hyperglycaemia, differed from the previously reported patients who only had non-autoimmune, insulinopenic diabetes not associated with ketoacidosis.

The combination of fasting hypoglycaemia and postprandial hyperglycaemia is well known to occur in patients with congenital hyperinsulinaemia (HI).13 A blunted, late insulin response to a glucose load is part of the altered physiology, at least in patients with ABCC8 mutations, the most common form of HI.14 The similarity to HI is also demonstrated in the glucose levels obtained in the formal OGTT performed in patient II-3 at 9 years of age (figure 2).14 Transition from recurrent hyperinsulinaemic hypoglycaemia to hypoinsulinaemic diabetes was also reported in HI15 ,16 perhaps as a consequence of the accelerated β cell apoptosis in these patients.17 Indeed, TRMT10 silencing by inhibition of specific RNA transcription in β cells was associated with accelerated apoptosis.12 The monitoring of glucose levels from birth in the 13th child in the family clearly shows that hypoglycaemia is absent in infancy at least until 10 months of age. This greatly reduces the likelihood that neonatal hypoglycaemia is the cause of the neurological deficiency; among children with HI, those who are neurologically damaged invariably had severe neonatal hypoglycaemia and were not microcephalic.18 Furthermore, although HI can present beyond infancy16 ,19 and hyperinsulism is present in many developmental syndromes,20 they cannot explain much of the abnormality in glucose metabolism in our patients . Clearly, other mechanisms play a role as exemplified by the hypoinsulinaemic ketotic hypoglycaemia documented in patient II-3 (perhaps due to liver involvement). Yet another mechanism, that is, insulin resistance, suggested by high fasting insulin and low-normal fasting glucose in patient II-3 could underlie the non-autoimmune insulinopenic diabetes reported in older patients with another inactivating TRMT10 mutation.12

All in all, we propose that inactivating mutations in the TRMT10A gene produce a syndrome associated with a multi-faceted defect of glucose metabolism in which transition occurs from hypoglycaemic episodes to diabetes, the exact mechanism of which remains to be elucidated.

PI TWB .25W Vertebrate genomes, including human, encode three homologues of yeast TRM10, the cytoplasmic TRMT10A and TRMT10B and the mitochondrial TRMT10C. It was previously proposed that the reason for the presence of two cytoplasmic tRNA m1G9 methyltransferases is related to different developmental or tissue-specific functions.8 Our patients had selective involvement of the β cells and the brain. TRMT10A protein was previously shown to be abundant in brain and β cells in humans12 and the absence of symptoms in other tissues in the six reported patients suggests that TRMT10B has overlapping activity in many organs, but cannot compensate for the lack of functional TRMT10A in brain, β cells and perhaps liver. The profound loss of enzymatic activity associated with the G206R alteration suggests that there is unlikely to be significant residual activity of TRMT10A in the patient cells that could explain the differences between the glucose metabolic defects observed in this study compared with the nonsense mutation. However, this remains to be demonstrated experimentally through analysis of tRNA modification levels. The loss of N-1 methylation at position 9 (m1A9 modification catalysed by another human ortholog, TRMT10C) is associated with dramatic structural defects in at least one human mitochondrial tRNA, and therefore it is possible that similar defects in tRNA structure caused by the TRMT10A mutation contribute to the underlying molecular cause of the human disease.21 ,22 Moreover, while deletion of TRM10 in Saccharomyces cerevisiae does not cause obvious growth defects under normal conditions, suggesting that tRNAs are functional without the G9 modification, the yeast trm10Δ strain is hypersensitive to growth in the presence of 5-fluorouracil.23 This phenotype, which is exacerbated by growth at elevated temperature, is suggested to be due to general tRNA instability caused by loss of the m1G9 modification. Finally, an intriguing report of altered levels of nucleotide modifications in mouse small RNAs from diabetic liver compared with normal tissue suggests the possibility of a more general connection between glucose metabolism and modified nucleotides.24 Nonetheless, the precise molecular function of m1G9 modification on any nuclear-encoded tRNA has not been conclusively demonstrated and remains to be investigated.

In summary, TRMT10A catalytic deficiency appears to cause a syndrome presenting with intellectual disability, microcephaly and delayed puberty. These features are associated with an unusual form of impaired glucose metabolism presenting in early to mid-childhood with hypoglycaemia. Later, although insulin secretion decreases, hypoglycaemia may persist due to other metabolic causes. Then, in adolescence and early adulthood, non-autoimmune insulinopenic diabetes becomes evident.


Dr Raphy Singer is acknowledged for his assistance with the collection of clinical data and procurement of blood samples.



  • or

  • Dr Jane E Jackman, Department of Chemistry and Biochemistry, Ohio State Biochemistry Program and Center for RNA Biology, The Ohio State University, Columbus, Ohio, 43210, USA

  • Contributors DG, JEJ and OE conceived and designed the experiments, analysed the data and wrote the paper; AK, BY and AS performed the experiments; DG undertook patient management, collection of samples and delineation of the phenotype.

  • Competing interests None.

  • Patient consent Obtained.

  • Ethics approval The ethical committees of Hadassah Medical Center and the Israeli Ministry of Health.

  • Provenance and peer review Not commissioned; externally peer reviewed.

  • Data sharing statement FASTQ files (8 Gb) are available on request.