A novel class of small RNAs: tRNA-derived RNA fragments (tRFs)

Abstract

New types of small RNAs distinct from microRNAs (miRNAs) are progressively being discovered in various organisms. In order to discover such novel small RNAs, a library of 17- to 26-base-long RNAs was created from prostate cancer cell lines and sequenced by ultra-high-throughput sequencing. A significant number of the sequences are derived from precise processing at the 5' or 3' end of mature or precursor tRNAs to form three series of tRFs (tRNA-derived RNA fragments): the tRF-5, tRF-3, and tRF-1 series. These sequences constitute a class of short RNAs that are second most abundant to miRNAs. Northern hybridization, quantitative RT-PCR, and splinted ligation assays independently measured the levels of at least 17 tRFs. To demonstrate the biological importance of tRFs, we further investigated tRF-1001, derived from the 3' end of a Ser-TGA tRNA precursor transcript that is not retained in the mature tRNA. tRF-1001 is expressed highly in a wide range of cancer cell lines but much less in tissues, and its expression in cell lines was tightly correlated with cell proliferation. siRNA-mediated knockdown of tRF-1001 impaired cell proliferation with the specific accumulation of cells in G2, phenotypes that were reversed specifically by cointroducing a synthetic 2'-O-methyl tRF-1001 oligoribonucleotide resistant to the siRNA. tRF-1001 is generated in the cytoplasm by tRNA 3'-endonuclease ELAC2, a prostate cancer susceptibility gene. Our data suggest that tRFs are not random by-products of tRNA degradation or biogenesis, but an abundant and novel class of short RNAs with precise sequence structure that have specific expression patterns and specific biological roles.

Figure 1.

Pipeline of analyses of 454 deep sequencing data. A schematic of the 454 sequencing data analyses is depicted.

Figure 2.

Classification and nomenclature of tRFs that were revealed by analysis of highly abundant nmsRNAs. (A) Identification and classification of 30 nmsRNA sequences that were cloned abundantly. The nmsRNAs are grouped by their identity in the pie chart. (unmappable) nmsRNAs unidentifiable in the human genome sequence or the human EST database, probably due to as yet unreported splicing or sequence editing events. (B,C) tRNA-related small RNAs (tRFs) are further classified by their relative location in the tRNA pre-tRNA. (B) Diagram of tRFs aligned at a tRNA locus (see C for actual sequence alignment). (Wavy line) The genomic sequence for pre-tRNA ending with an oligo-dT stretch, the 3′-termination signal of RNA polymerase III. (Gray bar) A mature tRNA after 3′ trimming of the pre-tRNA and addition of the CCA that is not present in the genomic sequence. (Black bars) Three groups of tRFs (tRF-5, tRF-3, and tRF-1 series) are aligned with the pre-tRNA or mature tRNA. tRFs are numbered, with the first digit indicating the group and the other digits forming a serial number in that group. (C) Each of the 17 tRF sequences in A (bold letters; see also Table 1) is aligned to the relative position (see also B) in the corresponding mature tRNA (shaded capital letters) or the flanking genomic regions (lowercase letters).

Figure 3.

Comprehensive analyses of tRFs: fraction of tRFs in nmsRNAs and characterization of cleavage sites for tRF-5 and tRF-3 series. (A) Among nmsRNAs that were cloned more than five times and mappable in the human genome, tRNA-related sequences (defined as sequences matching anywhere in mature tRNA sequence plus the flanking 25 nt at both ends) were counted and depicted in a pie chart shown at left. In the other two pie charts, tRNA-related sequences were further sorted into the three tRFs (defined and described in the text and Fig. 2B,C) or the rest (“nonspecific”). (B) Nucleotide composition at cleavage sites to generate tRF-5 and tRF-3 series. Nucleotides at a given position (numbered in the diagram) in tRFs were counted and normalized to the nucleotide counts of the relative positions of all 622 tRNAs (see also the legend for Supplemental Table S1). The height of each letter (A, C, G, and U) of stacked columns is proportional to the relative frequency of the nucleotide (see Supplemental Table S1 for actual values).

Figure 4.

tRF-1001 is highly expressed in proliferating cancer cells. (A, top panel) Northern hybridization of tRF-1001. (Bottom panel) The same amounts of RNA were run in a separate gel for EtBr staining. Source of tissue RNAs is described in Kim et al. (2006). tRF-1001 (19-mer) is indicated by an arrowhead on the left. The positions of 18- and 24-mer synthetic oligoribonucleotides (arrowheads on the right), along with a 10-base-pair DNA ladder, are shown as molecular size markers. (B) Measurement of tRF-1001 and its pre-tRNA upon serum depletion in DU145 cell line. At 72 h after serum starvation (see the Supplemental Material), DU145 cells were replenished with 10% FBS medium (“re-add”). Splinted ligation assay of tRF-1001 (top panel) and EtBr staining of total RNA for equal loading (bottom panel).

Figure 5.

The impaired growth of HCT116 cells upon siRNA-mediated knockdown of tRF-1001 is rescued by Me-tRF-1001. (A) The sequences of siRNA duplexes are shown in alignment with the pre-tRNA sequence of tRF-1001. si-tRF1001 was designed against the underlined region of pre-tRNA, encompassing tRF-1001 (capital letters) and the mature tRNA (lowercase letters). Arrowheads indicate expected sites in the pre-tRNA that will be cleaved by the si-tRF1001 (Elbashir et al. 2001b). (B) MTT and BrdU incorporation assays performed after two rounds of transfection (days 0 and 2) of siGL2 (plain bar), si-tRF1001 (black bar), and si-tRF1001 with Me-GL2 (light-gray bar) or Me-tRF-1001 (dark-gray bar). (Left panel) A scan of wells after MTT assay at day 4. (Right panel) BrdU incorporation normalized to MTT value at days 3 and 4. Average and standard deviation from triplicates are shown. The BrdU/MTT for cells transfected with siGL2 is set as 1. (C) FACS of HCT116 cells transfected (described in B), harvested at day 4, and stained with propidium iodide for DNA content. Cells in G1, S, or G2 were quantitated by ModFit LT for Mac (version 3.2.1). (Right panels) Change in percentage of cells in each phase of the cell cycle between the indicated two samples. (D) MTT assays upon transfection of indicated siRNAs and Me-oligos. After two rounds (days 0 and 1) of transfection of indicated oligonucleotides, MTT assays were performed at day 3. Average and standard deviation from tetraplicates are shown. The value of siGL2 plus Me-GL2 is set as 1. (E) MTT assays upon cotransfection of tRF-1001 mutants. The sequences of Me-tRF-1001 M1 and M2 are 5′-GCCUAGGGUGCUCUUAUUU-3′ and 5′-GAAGCGGGUGCUCAAUAUU-3′ (mutated portion italicized), respectively. All other descriptions are same as in D.

Figure 6.

tRF-1001 is generated by ELAC2 in the cytoplasm. (A) DU145 cells were transfected with an siRNA against ELAC2. (Top two panels) Knockdown of ELAC2 was confirmed by Western blot, with β-actin as a loading control. tRF-1001 and its pre-tRNA was measured by splinted ligation assays (middle two panels), with EtBr staining of total RNA (bottom panel) as a loading control. (B) Total, nuclear, and cytoplasmic RNAs from LNCaP cells were subjected to Northern hybridization of tRF-1001, snoU38 as a nuclear marker, and 5S rRNA and EtBr staining of total RNA as a loading control. (C) Measurement of tRF-1001 in total, nuclear, and cytoplasmic RNAs from DU145 cells. EtBr staining of total RNA is shown as both quantity and quality control (compare EtBr staining pattern of fractionated RNAs in B). (D) A model for biogenesis and regulation of tRF-1001. See the text for details.