BCDIN3D regulates tRNAHis 3' fragment processing

Abstract

5' ends are important for determining the fate of RNA molecules. BCDIN3D is an RNA phospho-methyltransferase that methylates the 5' monophosphate of specific RNAs. In order to gain new insights into the molecular function of BCDIN3D, we performed an unbiased analysis of its interacting RNAs by Thermostable Group II Intron Reverse Transcriptase coupled to next generation sequencing (TGIRT-seq). Our analyses showed that BCDIN3D interacts with full-length phospho-methylated tRNAHis and miR-4454. Interestingly, we found that miR-4454 is not synthesized from its annotated genomic locus, which is a primer-binding site for an endogenous retrovirus, but rather by Dicer cleavage of mature tRNAHis. Sequence analysis revealed that miR-4454 is identical to the 3' end of tRNAHis. Moreover, we were able to generate this 'miRNA' in vitro through incubation of mature tRNAHis with Dicer. As found previously for several pre-miRNAs, a 5'P-tRNAHis appears to be a better substrate for Dicer cleavage than a phospho-methylated tRNAHis. Moreover, tRNAHis 3'-fragment/'miR-4454' levels increase in cells depleted for BCDIN3D. Altogether, our results show that in addition to microRNAs, BCDIN3D regulates tRNAHis 3'-fragment processing without negatively affecting tRNAHis's canonical function of aminoacylation.

Fig 1. BCDIN3D stably interacts with mature tRNA^His.

A. RNAs purified from HeLa-S3-Flp-In and HeLa-S3-Flp-In-BCDIN3Df FLAG eluates were analyzed on a denaturing 15% polyacrylamide/urea gel stained with silver. The asterisks indicate non-specific RNAs. A 10 bp DNA Ladder was run in an adjacent lane and the position of its 100 bp band is indicated. Please note that the migration of the 10 bp ladder is offset by 30 to 50 nt compared to RNA. B. Schematic of TGIRT-seq protocol used to generate the next generation sequencing libraries. C. Results of the TGIRT-seq libraries from BCDIN3Df-interacting small RNAs (reads > 50nt) from two biological repeats. The bio-informatic pipeline is schematized in S1 Fig, and full results of the two biological repeats are shown in S1 Table. The sequences of BCDIN3D interacting RNAs >50nt mapping to tdbD00005717_His_GTG (tRNA^His) from one of the two biological repeats are represented in a WebLogo format (top) and in an IGV plot representation (bottom) that includes coverage plots to the reference sequence. The numbers of reads are indicated to the left of the coverage plot in brackets. Nucleotides in reads matching nucleotides in the reference are colored in gray, while mismatched nucleotides are color coded by nucleotide (A, green; C, blue; G, brown; and T, red). The nontemplated T residues at the 5’ end of tRNA^His correspond to nontemplated A nucleotide addition by TGIRT at cDNAs 3′ ends. Highlighted are the 5’ G_-1 and the 3’ CCA tail characteristic of mature tRNA^His, as well as the sites of m¹G₃₇ and m¹A₅₇ modifications that lead to the distinctive nucleotide misincorporation by TGIRT. D. Validation of the TGIRT-seq results with northern blotting using tRNA^His northern blot probe #1, the target position of which is shown in cyan in Fig 1C. The Northern blot with the U6 probe is to show equal background RNA in Control and BCDIN3Df FLAG eluates. E. BCDIN3D prefers to methylate mature tRNA^His-5’P containing the 5’ G_-1 residue. The indicated synthetic tRNA^His RNAs were in vitro methylated with BCDIN3D using radioactive [³H]-SAM as methyl group donor. The bottom panel shows the SYBR Gold stained gel that was used for the autoradiography in the top panels, while the graph shows the scintillation counts of C[³H]₃ incorporated into the RNA. F. THG1L prefers to guanylate tRNA^His without the 5’ G_-1 residue. tRNA^His-5’P with or without the G_-1 residue were in vitro guanylated with recombinant THG1L using radioactive [³²P]α-GTP. The bottom panel shows the SYBR Gold stained gel that was used for the autoradiography in the top panel.

Fig 2. tRNA^His interacting with BCDIN3D is phospho-methylated.

A. Synthetic pre-miR-145 with either [5’-OH], [5’-P], [5’-Pme1] or [5’-Pme2] ends was treated with Antarctic Phosphatase. Please note that pre-miR-145-5’-OH (cyan *) migrates faster than pre-miR-145 with [5’-P], [5’-Pme1] or [5’-Pme2] ends (red *) on a denaturing 15% polyacrylamide/urea gel. B. Synthetic pre-miR-145 [5’-P] was in vitro methylated with BCDIN3D using radioactive [³H]-SAM as methyl group donor prior to treatment with Antarctic Phosphatase. The left panel shows the ethidium bromide stained gel that was used for the autoradiography in the right panel, which shows only the in vitro methylated RNAs. Please note that only a small fraction of 5’P pre-miR-145 is dimethylated by BCDIN3D in vitro, explaining why AP treatment induces a gel shift in the left panel, which shows the total pre-miR-145, but not in right panel, which shows only the methylated pre-miR-145. C. Schematic showing the 5’ ends and the effect of Antarctic Phosphatase (AP) and Polynucleotide Kinase (PNK). D. Synthetic tRNA^His-5’-P was in vitro methylated with BCDIN3D using radioactive [³H]-SAM as methyl group donor prior to treatment with Antarctic Phosphatase. The left panel shows the ethidium bromide stained gel that was used for the autoradiography in the right panel. Please note that tRNA^His-5’-OH (cyan *) migrates slower than tRNA^His with [5’-P], [5’-Pme1] or [5’-Pme2] (red *) ends. E. RNAs purified from HeLa-S3-Flp-In and HeLa-S3-Flp-In-BCDIN3Df FLAG eluates were treated with mock, AP or PNK, and separated on a denaturing 15% polyacrylamide/urea gel and probed with the tRNA^His northern blot probe #1. Neither the treatment with AP nor T4 PNK shifted the migration of tRNA^His (Fig 1C). The treatment with PNK is a control to verify that the 5’ end is not 5’OH in BCDIN3Df-interacting tRNA^His.

Fig 3. BCDIN3D interacts with a miRNA that may be a tRNA^His 3’ fragment.

A. Results of the TGIRT-seq libraries from BCDIN3Df-interacting small RNAs (< 50nt) from the same two biological repeats as in Fig 1C. Full results are shown in S3 Table. B. Number of reads < 50nt mapping to the top most abundant RNAs in BCDIN3Df FLAG eluates. Shown are means ±SD of raw reads from two biological replicates. tRNA^His (tdbD00005717_His_GTG) and hsa-miR-4454 are the only two top 10 RNAs specifically enriched in BCDIN3Df versus Control FLAG eluates. C. Northern blot analysis of Control and BCDIN3Df FLAG eluates RNAs with tRNA^His probes #1 and #2. Note that the ladder is the ss20 ssDNA Ladder and its migration is offset by 10–20 nt compared to RNA. D. Shown are hsa-miR-4454 reads from BCDIN3Df FLAG eluates from two biological repeats in IGV plot representation. Reads are down-sampled to 100 per 50 bp window. Coverage plots are above read alignments to the hsa-miR-4454 reference sequence. The numbers of reads are indicated to the left of the coverage plot in brackets. Nucleotides in reads matching nucleotides in the reference are colored in gray, while mismatched nucleotides are color coded by nucleotide (A, green; C, blue; G, brown; and T, red). Coverage plot of the 3’ end of tRNA^His is also shown. The G4 position of the hsa-miR-4454 stem loop aligns to the A57 residue of tRNA^His and shows the same nucleotide polymorphism, dominated by T and G that is characteristic of TGIRT-III mis-incorporation at m1A. Please note that A57 in tRNA^His corresponds to A58 in other tRNAs due to a skipped position in the variable loop in tRNA^His.

Fig 4. BCDIN3D regulates tRNA^His 3’ fragment processing.

A. Dicer assay with 5’P and 5’Pme2 tRNA^His using 0, 1, 2 and 4 μL of human Dicer at 1 μg/μL. Note that the ladder is the ss20 ssDNA Ladder and its migration is offset by 10–20 nt compared to RNA. The graph on the right shows the quantification by ImageQuant of the indicated bands. B. RNAs from two independent in vitro Dicer assays using mock or 2 μl of human Dicer at 1 μg/μl and 20 pmol of tRNA^His-5’P or -5’Pme2 were sequenced by TGIRT-seq. The bars represent mean ± SD (n = 2) of the ratio of normalized full length tRNA^His reads (>73 nt) over the mock control (**, p-value of ~0.0095). C. The graph represents mean ± SD (n = 2) of the ratio of normalized tRNA^His reads ending at the indicated position. The deduced Dicer cuts are indicated on the diagram of tRNA^His and the sequence of tRNA^His aligned to hsa-miR-4454. The red and grey arrows indicate the position of Dicer cuts that are sensitive to tRNA^His phospho-methylation. D. The graph on the left shows the RTqPCR analysis of the BCDIN3D mRNA normalized to ALAS1 and B2M, and the images on the right show the western blot analysis of the BCDIN3D protein from MDA-MB-231shNC and shBCDIN3D (shB3D) cells. E. Average ± SEM (n = 2) of tRNA^His 3’ fragment reads normalized to total mapped reads relative to shNC. F. Analysis of tRNA^His aminoacylation in MDA-MB-231shNC and shBCDIN3D cells. 1 μg of RNAs purified under acidic conditions, which preserve the aminoacyl-tRNA ester bond, was analyzed by acidic gel and northern blot. As a control, 1 μg of RNA was deacylated in vitro prior to gel migration. The slower migrating band corresponds to the aminoacylated tRNA^His (charged), while the faster migrating band corresponds to the un-aminoacylated tRNA^His (uncharged). G. Quantification of acidic northern blot with tRNA^His probe from 2 biological replicates, including the acidic northern blot in Fig 4F. The graph shows average ± SEM (n = 2). The total levels of tRNA^His are normalized to shNC, and the aminoacylation ratios of tRNA^His are calculated as the ratio of the signal of the aminoacylated band over the sum of both bands.