Targeted sequencing reveals expanded genetic diversity of human transfer RNAs

Abstract

Transfer RNAs are required to translate genetic information into proteins as well as regulate other cellular processes. Nucleotide changes in tRNAs can result in loss or gain of function that impact the composition and fidelity of the proteome. Despite links between tRNA variation and disease, the importance of cytoplasmic tRNA variation has been overlooked. Using a custom capture panel, we sequenced 605 human tRNA-encoding genes from 84 individuals. We developed a bioinformatic pipeline that allows more accurate tRNA read mapping and identifies multiple polymorphisms occurring within the same variant. Our analysis identified 522 unique tRNA-encoding sequences that differed from the reference genome from 84 individuals. Each individual had ~66 tRNA variants including nine variants found in less than 5% of our sample group. Variants were identified throughout the tRNA structure with 17% predicted to enhance function. Eighteen anticodon mutants were identified including potentially mistranslating tRNAs; e.g., a tRNA^Ser that decodes Phe codons. Similar engineered tRNA variants were previously shown to inhibit cell growth, increase apoptosis and induce the unfolded protein response in mammalian cell cultures and chick embryos. Our analysis shows that human tRNA variation has been underestimated. We conclude that the large number of tRNA genes provides a buffer enabling the emergence of variants, some of which could contribute to disease.

Keywords: human tRNA variation; tRNA biology; tRNA capture panel; tRNA-encoding genes; translation.

Figure 1.

tRNA structure. (a) tRNAs are represented as cloverleaf structures in two dimensions. (b) In three dimensions, tRNAs fold into an L-shape stabilized by intramolecular base-pairing shown here by the tRNA^Phe structure [PDB: 1HEZ; [20]]. In both diagrams, the tRNA structural elements are colored: acceptor stem (green), dihydrouridine (D)-arm (purple), anticodon stem (light blue), anticodon (bases 34, 35, 36 in dark blue), variable arm (orange), T-arm (yellow) and the discriminator base (red).

Figure 2.

Workflow for sequencing and mapping tRNA genes, followed by variant calling, annotation and functional predictions used in this study.

Figure 3.

tRNA variation in individuals. Number of tRNA variants per person in males and females for (a) total variants as compared to the reference genome and (b) variants with allele frequencies less than 25% or (c) less than 5% in our sample dataset. The mean number of variants in each set is indicated (black bar). (d) Heat map of the tRNA variation profile for each individual. On the x-axis, males are grouped on the left and females on the right. Each row on the y-axis represents an individual tRNA locus or groups of tRNA where reads could not be uniquely assigned. Groups of tRNAs are denoted by black bars on the right side of the heatmap. The tRNA genes were hierarchically clustered using complete linkage and Euclidean distance. Each tRNA is labelled as either high confidence (orange) or low confidence (purple). tRNAs genes where variation was not observed are not included.

Figure 4.

Distribution of tRNA-encoding genes with variation. Break down of tRNA genes in which we observed variation by (a) isoacceptor and (b) high confidence and low confidence annotation. In (a), ‘Und’ represents tRNAs whose identity and anticodon are undefined in the tRNA database [18]. (c) Variation accumulation curve plotted using the R package ‘vegan’ [92] with 100 subsamplings without replacement for each grouping of individuals on the x-axis. The red line represents the mean number of tRNA loci with mutations for each grouping size of individuals and the grey boundaries represent the standard deviation.

Figure 5.

tRNA variation is distributed throughout the tRNA structure. (a) The location of all single nucleotide variants in high confidence tRNAs was mapped onto the canonical two-dimensional tRNA structure, using the canonical numbering [19]. Nucleotides colored darker blue have the least number of variants, white nucleotides have an intermediate number and red nucleotides have the most variants. Insertions, deletions and variants in the extended variable arm of serine and leucine tRNAs or within introns were not included. (b) The location of all single nucleotide variants from (a) were mapped onto the three-dimensional tRNA structure. Coloring is the same as in (a). (c) The number of variants at each position was plotted for high confidence tRNAs for alleles occurring in less than 5% of our sample population (red dotted line) and for all variant alleles identified in this study (black solid line).

Figure 6.

Predicted effect of tRNA variation on expression and function. The (a) EufindtRNA score and (b) Infernal score for each variant tRNA (red circle) and its reference tRNA (black circle) was computed using tRNAscan-SE [36] and rank ordered from lowest to highest reference tRNA score.

Figure 7.

tRNA variants with multiple nucleotide changes. The Infernal score was calculated for high confidence tRNAs containing multiple nucleotide changes. Variants that decreased score are shown on the left, variants whose scores were neutral and did not change by more than 1 bit are in the middle and variants whose scores increased are on the right for variants containing (a) two or (b) three polymorphisms in the same allele. Each set of letters represent the position of single nucleotide changes within a single tRNA allele. Variants in the intron or extended variable arm are not shown.