5' tRNA halves are highly expressed in the primate hippocampus and might sequence-specifically regulate gene expression

Abstract

Fragments of mature tRNAs have long been considered as mere degradation products without physiological function. However, recent reports show that tRNA-derived small RNAs (tsRNAs) play prominent roles in diverse cellular processes across a wide spectrum of species. Contrasting the situation in other small RNA pathways the mechanisms behind these effects appear more diverse, more complex, and are generally less well understood. In addition, surprisingly little is known about the expression profiles of tsRNAs across different tissues and species. Here, we provide an initial overview of tsRNA expression in different species and tissues, revealing very high levels of 5' tRNA halves (5' tRHs) particularly in the primate hippocampus. We further modulated the regulation capacity of selected 5' tRHs in human cells by transfecting synthetic tsRNA mimics ("overexpression") or antisense-RNAs ("inhibition") and identified differentially expressed transcripts based on RNA-seq. We then used a novel k-mer mapping approach to dissect the underlying targeting rules, suggesting that 5' tRHs silence genes in a sequence-specific manner, while the most efficient target sites align to the mid-region of the 5' tRH and are located within the CDS or 3' UTR of the target. This amends previous observations that tsRNAs guide Argonaute proteins to silence their targets via a miRNA-like 5' seed match and suggests a yet unknown mechanism of regulation. Finally, our data suggest that some 5' tRHs that are also able to sequence-specifically stabilize mRNAs as up-regulated mRNAs are also significantly enriched for 5' tRH target sites.

Keywords: gene regulation; k-mer mapping; small noncoding RNAs; tRNA fragments; target identification; target prediction.

FIGURE 1.

Small RNA annotation of mapped reads from small RNA sequencing libraries of samples from the human hippocampus (left), frontal cortex (middle), and cerebellum (right). (A–C) Proportion of small RNA classes. (D–F) Proportion of tsRNA classes. (G–I) Proportion of 5′ tRH species. (J–L) Relative abundances of top 5′ tRH species in comparison to top miRNA species as a measure of reads per million (RPM).

FIGURE 2.

Phylogenetic tree and small RNA annotation of mapped reads from small RNA sequencing libraries of hippocampal samples from mouse, rat, pig, human, chimpanzee, and macaque.

FIGURE 3.

(A,B) qPCR quantified log₂ fold change of potential 5′ tRH targets in 5′ tRH overexpression HEK293T cells compared to control cells. The selected transcripts were predicted by miRanda (red) and piRanha (orange) to be targets of the 5′ tRNA-halves Glu-CTC (A) or Gly-GCC (B). The housekeeping genes ACTB and RPS18 were used as normalizers. Note that normalization by ACTB is not suitable in case of the 5′ tRH-Glu-CTC overexpression (A), as the expression of ACTB is up-regulated in this condition (also revealed by RNA-seq; see C). (C,D) Volcano plot of differential expression analysis for protein-coding genes of 5′ tRH overexpression and control HEK293T cells (blue, adjusted P-value <0.01). The 10 previously tested putative target genes are highlighted in red (miRanda prediction) and orange (piRanda prediction). For the 5′ tRH-Glu-CTC analysis (C), the housekeeping genes ACTB and RPS18 are additionally highlighted in pink, showing that ACTB gets up-regulated upon 5′ tRH-Glu-CTC overexpression, while the RPS18 expression is not affected. (E,F) Cumulative fraction plots for log₂-fold-change values of genes that were identified as potential targets of the 5′ tRHs Glu-CTC or Gly-GCC by miRanda (red; threshold: miRanda-score < −80) or piRanha (orange; threshold: piRanha-score < −30) and of genes not predicted as targets by the respective algorithm (black).

FIGURE 4.

Graphical overview of identification of tsRNA targeting rules via a k-mer mapping approach.

FIGURE 5.

k-mer analysis to elucidate the targeting rules of 5′ tRH-mediated transcript regulation in HEK293T cells. The y-axis refers to the fraction of up- or down-regulated genes. Blue dots refer to genes that have a corresponding k-mer alignment, red dots refer to genes that do not have a corresponding k-mer alignment. Black dots refer to the overall fraction of genes that align to the corresponding k-mer (declines with growing k-mer size). The numbers in the gray boxes above each plot indicate the respective k-mer length in nucleotides. That is, dots within the first column (with k-mer length = 5 nt) all refer to genes that align to k-mers with a length of 5 nt. Further, dots at the very left of a column refer to genes that align to k-mers from the very 5′ end of the parent tRH, while dots at the very right of a column refer to genes that align to k-mers from the very 3′ end of the parent tRH. Dots in between correspond to genes that align to k-mers derived from a corresponding position inside the parent tRH.

FIGURE 6.

(A) Volcano plot of differential expression analysis for protein-coding genes of 5′ tRH antisense-inhibition and control HepG2 cells (blue, adjusted P-value <0.01). (B) Venn diagram of significantly differentially expressed genes. Genes that are likewise regulated in HEK as in HepG2 cells are highlighted in gray. Genes that are inversely regulated in HEK overexpression and HepG2 inhibition cells are highlighted rosy (“potential perish targets”) and lutescent (“potential shelter targets”). (C) Log₂ fold change values of perish and shelter transcripts in the HEK overexpression and HepG2 antisense-inhibition experiments. (D,E) Analysis of thermodynamically favored alignments for the major transcript regions of “potential perish targets” or “potential shelter targets” with 5′ tRH-Glu-CTC.

FIGURE 7.

Target pattern analysis of published RNA sequencing data of fly S2 cells, where the 20 nt long 5′ tsRNA Glu-CTC, 5′ tsRNAs Asp-GTC or Lys-TTT was overexpressed by tsRNA mimic transfection (Luo et al. 2018). Plotted is the percentage of genes with (blue) or without (red) k-mer alignment that get down-regulated.