Copy number variation in tRNA isodecoder genes impairs mammalian development and balanced translation

Abstract

The number of tRNA isodecoders has increased dramatically in mammals, but the specific molecular and physiological reasons for this expansion remain elusive. To address this fundamental question we used CRISPR editing to knockout the seven-membered phenylalanine tRNA gene family in mice, both individually and combinatorially. Using ATAC-Seq, RNA-seq, ribo-profiling and proteomics we observed distinct molecular consequences of single tRNA deletions. We show that tRNA-Phe-1-1 is required for neuronal function and its loss is partially compensated by increased expression of other tRNAs but results in mistranslation. In contrast, the other tRNA-Phe isodecoder genes buffer the loss of each of the remaining six tRNA-Phe genes. In the tRNA-Phe gene family, the expression of at least six tRNA-Phe alleles is required for embryonic viability and tRNA-Phe-1-1 is most important for development and survival. Our results reveal that the multi-copy configuration of tRNA genes is required to buffer translation and viability in mammals.

Fig. 1. Evolutionary diversity and distribution of mammalian multigene tRNA families.

a Classes of tRNA familes. A tRNA isotype refers to a tRNA that is charged with one of the same 20 common amino acids, isoacceptor tRNAs are tRNAs that have distinct anticodons but are charged with the same amino acid, and isodecoders have the same anticodon but sequence differences in the rest of the tRNA body. b Number of tRNA isoacceptor and tRNA isodecoders in diverse species as listed by GtRNAdb. c GWAS associations in close proximity to human tRNA-Phe-GAA genes with number of significant GWAS hits (p < 0.05, one-way ANOVA) within up- and down-stream windows of ±50, ±100, ±250, ±500 and ±1000 bp of the tRNA gene. Individual disease annotations were summarised into disease categories. d Numbers of tRNA-Phe genes identified across six species of mammals from GtRNAdb. e Alignment of tRNA-Phe gene sequences from six mammalian species. Identical sequences within a species were collapsed and indicated by*. For example, mouse tRNA-Phe-1-1, 1-2, 1-3, 1-4 and 1-5 all have identical sequences and have been collapsed, represented in the figure as Mouse tRNA-Phe-1-*. Horizontal yellow bars highlight identical sequence clusters, vertical purple bars show variability at that position and pink squares indicate single or low represented sequence variations. Secondary tRNA structural information is annotated at the bottom of the alignment. f Chromatin accessability of the seven tRNA-Phe genes in brain and liver samples isolated from 5 control mice in brain and 4 control mice in liver, determined by ATAC-Seq. All values are means ± SD. g Occupancy of RNA polymerase III at the promoters of the seven tRNA-Phe genes determined by ChIP-Seq (ChIP-Atlas). tRNA-Phe-1-1 n = 52, 1-2 n = 24, 1-3 n = 54, 1-4 n = 40, 1-5 n = 40, 2-1 n = 6 and 3-1 n = 51. Values are means ± SD.

Fig. 2. Tissue-specific distribution of tRNA-Phe isodecoders.

a Schematic representing the seven mouse lines generated using a single CRISPR/Cas9 guide to knockout each of the tRNA-Phe genes in the mouse genome. The guide RNA was injected in mouse embryos to introduce deletions in each of the seven tRNA-Phe genes. Mice with tRNA-Phe gene deletions were identified by sequencing and individual tRNA-Phe gene deletions were backcrossed to wild-type mice to generate single tRNA-Phe gene deletions. The single tRNA-Phe gene deletion mice were bred for at least 10 generations before commencing molecular and phyiological experiments. Single tRNA-Phe gene deletions were interbred to generate multiple tRNA-Phe gene deletions. b The abundance of total tRNA-Phe was measured by northern blotting in brain, liver, spleen, kidney and heart, in control and tRNA-Phe knockout mice for each of the seven tRNA-Phe knockout lines in at least three independent biological experiments with similar results. 18S rRNA was used as a loading control, and one representative blot is shown for each tissue. All values presented in panel b are means ± SD of n = 3. *p < 0.05, **p < 0.01, ***p < 0.001, Student’s two-way t-test (p = 0.039 for brain tRNA-Phe-1-1, p = 0.026 for brain tRNA-Phe-3-1, p = 0.0017 for liver tRNA-Phe-1-1, p = 0.017 for spleen tRNA-Phe-1-1 and p = 0.00048 for spleen tRNA-Phe-1-2). c Body weight differences between control and tRNA-Phe knockout male mice at 10 weeks of age. All values presented in panel c are means ± SD of n = 4. **p < 0.01, Student’s two-way t-test (p = 0.0011 for tRNA-Phe-1-1). Photo illustrates the typical size differences between wild-type (WT) and tRNA-Phe-1-1 knockout mice. d Tissue weight-to-body weight ratio of brain, liver, spleen, kidney and heart in control and tRNA-Phe knockout male mice at 10 weeks of age, values are means ± SD (n = 4); Student’s two-way t-test (p = 0.001 for brain tRNA-Phe-1-1, p = 0.0055 for liver tRNA-Phe-1-1, p = 0.000061 for kidney tRNA-Phe-1-1, p = 0.032 for kidney tRNA-Phe-1-2, p = 0.005 for kidney tRNA-Phe-3-1, and p = 0.041 for heart tRNA-Phe-1-1).

Fig. 3. Transcriptome and proteome-wide molecular signatures of neurological defects in the absence of tRNA-Phe-1-1 genes in the brain.

Transcriptome-wide changes in brains a, and livers b, from tRNA-Phe-1-1^-/- knockout mice (n = 3) compared to controls (n = 3), summarised by biological process gene ontologies (GOs) determined by PANTHER and visualised using REVIGO. GO terms have been summarised and grouped by REVIGOs internal clustering algorithm. GO size represents the number of total genes in each specific ontology, the colour scale represents the degree of significance and parent GO terms are marked with a red outline. Brain c, and liver d, proteomes of tRNA-Phe-1-1^-/- knockout mice (n = 5) compared to controls (n = 5), summarised as biological process GOs determined by PANTHER and REVIGO and visualised with CirGO. GO terms have been summarised and grouped by REVIGO’s internal clustering algorithm. The pie-charts display the proportion of genes changing from each GO category as generated by CirGO with proportions indicated and top GO terms that are collapsed within each representative term are shown as a bar graph with log10(FDR) and gene set size for each ontology.

Fig. 4. Effect of tRNA-Phe gene loss on brain morphology and behaviour in 10-week old mice.

a, b Sagittal sections of the brain in control (Phe^+/+), tRNA-Phe-1-1^-/-, tRNA-Phe-2-1^-/- and tRNA-Phe-3-1^-/- mice (n = 7 of each genotype) were stained with toluidine blue and assessed for histological changes in the cortex a, Cortex (pia is at the top), labels in the first panel show layers I to VI and the white matter (WM). b Hippocampus (DG: dentate gyrus, CA: cornu ammonis (hippocampus) regions). c vermis of the cerebellum. d hemisphere of the cerebellum. e Immunostaining with calbindin in sections from the hemisphere of the cerebellum (paramedian: PM lobe shown) revealed a decrease in calbindin-positive Purkinje cells in the tRNA-Phe1-1 knockout mice compared to controls. All values are means ± SEM of n = 5 *p < 0.05, **p < 0.01, (p = 0.0034 for PM and p = 0.00066 for the copula of the pyramis: Cop), two-way, unpaired, Student’s t-test; ML molecular layer, PCL Purkinje cell layer (indicated with a white arrowhead), GCL granule cell layer, C cortex layers, S subiculum. All experiments shown in a–e were repeated independently up to seven times with similar results. f Heatmaps representing the layout of the open field testing area show the overall movement of control (Phe^+/+), tRNA-Phe-1-1^-/-, tRNA-Phe-2-1^-/- and tRNA-Phe-3-1^-/- mice (n = 7 of each genotype). Shading represents the number of times a box was entered over a 10-min testing period. g The distance travelled by the control (Phe^+/+), tRNA-Phe-1-1^-/-, tRNA-Phe-2-1^-/- and tRNA-Phe-3-1^-/- mice was measured using DeepLabCut (n = 7), values are means ± SD of n = 7 *p < 0.05, two-way Student’s t-test and one-way ANOVA (p = 0.025 for tRNA-Phe-1-1). h Duration of escape-oriented behaviour recorded over a 5 min testing period in control and knockout mice during the tail suspension test. All values are means ± SD of n = 7 *p < 0.05, two-way Student’s t-test and one-way ANOVA (p = 0.012 for tRNA-Phe-1-1).

Fig. 5. In the absence of tRNA-Phe-1-1, translation is compensated by increased expression of other tRNAs.

a ATAC-Seq was used to identify changes in chromatin accessibility (normalised counts) of the seven tRNA-Phe genes in control mice compared to tRNA-Phe-1-1^-/-, tRNA-Phe-2-1^-/- and tRNA-Phe-3-1^-/- knockout mice in brain. Results show the mean score for each set of replicates (n = 3 of each genotype) over tRNA-Phe gene regions (in different colour) for each of the three knockout mouse lines compared to control mice in brain; all values are means ± SEM. ***p < 0.001 (p = 0.000045 for tRNA-Phe-1-1 locus in the tRNA-Phe-1-1^-/- mice, and p = 0.0006 for tRNA-Phe-3-1 locus in the tRNA-Phe-3-1^-/- mice), Student’s two-way t-test. b Significant changes in chromatin accessibility of tRNA genes in brains of tRNA-Phe-1-1^-/- mice compared to controls, determined using ATAC-Seq; values are log₂ fold changes ±SEM (n = 3 of each genotype), and all the significant changes determined using DESeq2 are shown. c Northern blotting of altered tRNAs in brains of tRNA-Phe-1-1^-/- mice compared to controls (n = 3 of each genotype); values are means ± SD; Student’s two-way t-test (p = 0.001 for tRNA-Tyr, p = 0.0055 for tRNA-Asp, and p = 0.041 for tRNA-Phe). d ATAC-Seq tracks show changes in the accessibility of nucleosome-free, mono-, di- and trinucleosome-bound chromatin in the brain within the tRNA-Phe-1-1 locus of tRNA-Phe-1-1^-/- mice compared to controls. Coverage tracks are expressed in reads per million (RPM) and heatmap tracks show log₂ fold change.

Fig. 6. Amino acid misincorporation reduces the abundance of proteins with high phenylalanine content.

a Biological processes most affected in ATAC-Seq of tRNA-Phe-1-1^-/- mouse brains compared to controls, summarised using PANTHER and REVIGO. b Transcriptomic and proteomic changes in tRNA-Phe-1-1^-/- mouse brains. Red indicates significant (s < 0.01 determined by DESeq2, adj p < 0.05 determined using Spectronaut) log₂-fold increases compared to the control and blue indicates decreases. Numbers of transcripts or peptides changed are indicated. c Comparison of phenylalanine usage in proteins compared to their change in abundance in the brain proteomes of tRNA-Phe-1-1^-/- mice relative to controls. Percentage of phenylalanine content in each detected protein is plotted against average log₂ ratios± SD (grey) in brain proteomes from tRNA-Phe-1-1^-/- mice; calbindin and calretinin are highlighted in red. Spearman’s rank correlation coefficient and p-value are shown. d Calbindin and calretinin levels in brains of tRNA-Phe-1-1^-/- mice (n = 5), determined by Spectronaut (*adj p = 0.026, **adj p = 0.0053). e Immunoblotting of calretinin in tRNA-Phe-1-1^-/- and control brain homogenates (n = 3), quantitated relative to β-actin. p = 0.027, Student’s two-way t-test, values are means ± SD. f Comparison of the common, significantly changing proteins (adj p-value < 0.05) and transcripts (adj p-value < 0.05) in the brain tRNA-Phe-1-1^-/- mice relative to controls. g Amino acid content in proteins that are reduced in brains of tRNA-Phe-1-1^-/- mice identified by mass spectrometry. Amino acids that occur at a higher frequency in proteins reduced in tRNA-Phe-1-1^-/- mice are in red, amino acids with lower lower occurrence are in blue; S p = 0.00041, H p = 0.016, W p = 0.044, Y p = 0.0025, I p = 0.021, L p = 0.049, F p = 0.00022. Student’s two-way t-test was used for c, f, g. h Mistranslated peptides identified in the brain proteomes of tRNA-Phe-1-1^-/- mice and their ratio relative to their canonical peptides in WT and tRNA-Phe-1-1^-/- mice. i Metagene analysis of averaged ribo-profiling in control and tRNA-Phe-1-1^-/- mice (n = 3), aligned to Phe (UUU), Phe (UUC) codons and an unrelated Ile (AUC) codon shows a specific increase in aligned read counts upstream of Phe (UUU) codons, and indicates ribosomal stalling at the A, P and E sites in the brains of tRNA-Phe-1-1^-/- mice.

Fig. 7. Hierarchical requirements of tRNA-Phe genes for embryonic viability and development.

a Expected and observed Mendelian inheritance of gene loci homozygous for the deletion of up to four tRNA-Phe alleles including tRNA-Phe-1-1. Black denotes homozygous loss of an allele, grey heterozygous loss of one allele and white is indicative of two wild-type alleles. b Embryos with tRNA-Phe-1-1^-/- and three additional homozygous tRNA-Phe deletions were underdeveloped at day E10 when compared to controls. c Genotypes of breeding pairs set up to produce mice with multiple tRNA-Phe deletions, including either a heterozygous or homozygous tRNA-Phe-1-1 deletion (black represents homozygous loss of the allele, grey is heterozygous loss of one allele and white is indicative of two wild-type alleles), and percentage of expected live births. d Observed percentage of live births (where white is no animals observed with this phenotype). The number of embryos examined are shown on the right of the heatmap. e Stillborn mice born at varied stages of development, with an increasing number of tRNA-Phe knockout alleles correlating to an earlier developmental failure. f, Normal development of E10 embryos with tRNA-Phe-1-1^+/+ and tRNA-Phe-1-2^-/-, tRNA-Phe-1-3^-/-, tRNA-Phe-1-4^-/- and tRNA-Phe-1-5^-/- compared to controls. g Canonical amino acid incorporation takes place during translation in wild-type mice expressing tRNA-Phe genes that maintain chromatin openness. Deletion of the dominant tRNA-Phe-1-1 allele, causes chromatin condensation at the tRNA-Phe allele and leads to amino acid misincorporation that reduces the stability of specific proteins required for neurological function and development. Sequential deletion of seven or more tRNA-Phe alleles reduces embryo viability and survival of mice.