Molecular basis for t6A modification in human mitochondria

Abstract

N 6-Threonylcarbamoyladenosine (t6A) is a universal tRNA modification essential for translational accuracy and fidelity. In human mitochondria, YrdC synthesises an l-threonylcarbamoyl adenylate (TC-AMP) intermediate, and OSGEPL1 transfers the TC-moiety to five tRNAs, including human mitochondrial tRNAThr (hmtRNAThr). Mutation of hmtRNAs, YrdC and OSGEPL1, affecting efficient t6A modification, has been implicated in various human diseases. However, little is known about the tRNA recognition mechanism in t6A formation in human mitochondria. Herein, we showed that OSGEPL1 is a monomer and is unique in utilising C34 as an anti-determinant by studying the contributions of individual bases in the anticodon loop of hmtRNAThr to t6A modification. OSGEPL1 activity was greatly enhanced by introducing G38A in hmtRNAIle or the A28:U42 base pair in a chimeric tRNA containing the anticodon stem of hmtRNASer(AGY), suggesting that sequences of specific hmtRNAs are fine-tuned for different modification levels. Moreover, using purified OSGEPL1, we identified multiple acetylation sites, and OSGEPL1 activity was readily affected by acetylation via multiple mechanisms in vitro and in vivo. Collectively, we systematically elucidated the nucleotide requirement in the anticodon loop of hmtRNAs, and revealed mechanisms involving tRNA sequence optimisation and post-translational protein modification that determine t6A modification levels.

Figure 1.

OSGEPL1 is a monomer and does not interact with YrdC. (A) Determination of the molecular mass of OSGEPL1 based on protein standards and elution volumes. (B) OSGEPL1-FLAG and OSGEPL1-Myc were co-expressed in HEK293T cells, and OSGEPL1-Myc could not be pulled down by OSGEPL1-FLAG in a Co-IP assay. (C) OSGEPL1-FLAG could not be precipitated by OSGEPL1-Myc in a Co-IP assay. (D) Dimeric Qri7 was used as a positive control. Qri7-FLAG and Qri7-Myc were co-expressed in HEK293T cells, and Qri7-Myc could be pulled down by Qri7-FLAG in a Co-IP assay. (E) Dimeric hmThrRS was used as a positive control, and hmThrRS-Myc was pulled down by hmThrRS-FLAG, as expected. (F) OSGEPL1-FLAG and YrdC-HA were co-expressed in HEK293T cells, and YrdC-HA could not be pulled down by OSGEPL1-FLAG in a Co-IP assay. (G) OSGEPL1-FLAG could not be precipitated by YrdC-HA in a Co-IP assay.

Figure 2.

OSGEPL1 possesses ATPase activity. (A) A representative TLC image showing AMP formation via ATP hydrolysis in the absence or presence of FeCl₃. A reaction without adding enzyme was included as a negative control. (B) Quantification analysis of AMP formation by OSGEPL1 with (black filled circles) or without (red circles) FeCl₃. A reaction without addition of OSGEPL1 was included as a control (blue diamonds).

Figure 3.

The hmtRNA^Thr transcript is the best substrate of YrdC/OSGEPL1 in vitro. (A) Time-course curves of the modification of five hmtRNAs, hmtRNA^Asn (green triangles), hmtRNA^Ile (red squares), hmtRNA^Lys (blue diamonds), hmtRNA^Ser(AGY) (pink hexagons) and hmtRNA^Thr (black filled circles), by YrdC and OSGEPL1. (B) t⁶A modification at increasing concentrations of hmtRNA^Asn (10 μM, red filled squares; 20 μM, green filled triangles; 40 μM, blue filled inverted triangles and 80 μM, pink filled diamonds). A control without tRNA addition (no tRNA, orange circles) was included. (C) LC–MS/MS analysis of the digestion product of the hmtRNA^Thr transcript in the absence and presence of standard t⁶A or modified hmtRNA^Thr transcripts by YrdC/OSGEPL1 or Sua5/Qri7.

Figure 4.

Nucleotide requirements in the anticodon loop. (A) Mutagenesis of hmtRNA^Thr at the non-Watson-Crick base pair (A29-C41; left), and mutagenesis of hmtRNA^Met at the anticodon loop base (C34 and C38; right) are coloured red. (B) t⁶A modification levels of hmtRNA^Thr (black filled circles) and its mutants C32A (red filled squares), C32U (magenta filled triangles) and C32G (blue filled inverted triangles). (C) t⁶A modification levels of hmtRNA^Thr (black filled circles) and mutants U33A (red filled squares), U33C (cyanine filled triangles) and U33G (blue filled inverted triangles). (D) t⁶A modification levels of hmtRNA^Thr (black filled circles) and mutants U34A (red filled squares), U34C (cyanine filled triangles) and U34G (blue filled inverted triangles). (E) t⁶A modification levels of hmtRNA^Thr (black filled circles) and mutants G35A (red filled squares), G35C (cyanine filled triangles) and G35U (magenta filled inverted triangles). (F) t⁶A modification levels of hmtRNA^Met (black filled circles) and mutants C34U (red filled squares), C38A (green filled triangles) and C34U/ C38A (magenta filled inverted triangles).

Figure 5.

C34 is not an anti-determinant for the yeast t⁶A modification machinery. (A) t⁶A modification levels of ScmtRNA^Arg(UCU) (red filled circles) and ScmtRNA^Arg(UCU)-U34C (black filled squares) determined with Sua5/Qri7. (B) t⁶A modification levels of SctRNA^Thr(CGU) (red filled triangles) and SctRNA^Thr(AGU) (green filled circles) determined with Sua5/ScKEOPS.

Figure 6.

Sequences of hmtRNA^Ile and hmtRNA^Ser(AGY) are fine-tuned for t⁶A modification. (A) t⁶A modification levels of hmtRNA^Ile (black filled circles) and mutants G38A (red filled squares), G38C (cyanine filled triangles) and G38U (magenta filled inverted triangles) by YrdC/OSGEPL1. (B) t⁶A modification levels of hmtRNA^Ser(AGY) (black filled circles) and mutants A31G (blue filled squares), C39U (magenta filled triangles), A42U (red filled inverted triangles) and C39U/A42U (green filled diamonds) by YrdC/OSGEPL1. (C) t⁶A modification levels of hmtRNA^Thr/Ser (black filled circles) and mutants A31G (blue filled squares), C39U (cyanine filled triangles), A42U (magenta filled triangles) and C39U/A42U (dark green diamonds) by YrdC/OSGEPL1.

Figure 7.

MTS-deleted OSGEPL1 can rescue loss of Kae1 in vivo. (A) Genes encoding Kae1, Qri7, Qri7-ΔMTS, OSGEPL1, OSGEPL1-ΔMTS and OSGEP were transformed into ScΔKae1, transformants were spread on SD/Leu⁻ or SD/Leu⁻/5-FOA plates, and the growth phenotype was observed. Kae1 and p425TEF empty vectors were used as positive and negative controls, respectively. (B) Yeast growth curves were determined after 5-FOA selection in SD/Leu⁻ liquid culture at an initial cell density (OD₆₀₀) of 0.03.

Figure 8.

The activity of OSGEPL1 is influenced by post-translational modification. Higher energy collision-induced dissociation (HCD) MS/MS spectra were recorded for (A) the [M+2H]²⁺ ion at m/z 821.9313 for human OSGEPL1 peptide SLDIAPGDMLDKVAR harbouring one acetylated site (Lys203), and (B) the [M+4H]⁴⁺ ion at m/z 534.5298 for the human OSGEPL1 peptide AADIAATVQHTMACHLVKR harbouring one acetylated site (Lys299). Predicted b- and y-type ions (not all) are listed above and below the peptide sequences, respectively. Matched ions are labelled in the spectra. (C) Genes encoding Kae1, OSGEPL1-ΔMTS, -K74Q, -K140Q, -K203Q, -K230Q, -K240Q and -K299Q were transformed into ScΔKae1, transformants were initially cultured in SD/Leu⁻ liquid medium, spread on SD/Leu⁻ or SD/Leu⁻/5-FOA plates, and the growth phenotype was observed on two plates treated with the same 10-fold diluted concentrations (initial OD₆₀₀ = 1.0) as indicated. Kae1 and p425TEF empty vector were used as positive and negative controls, respectively. (D) Western blotting analysis was performed with yeast extracts before 5-FOA selection. (E) t⁶A modification activities of OSGEPL1-ΔMTS (brown filled circles), K74Q (red filled squares), K203Q (green filled inverted triangles), K230Q (blue filled diamonds), K240Q (violet filled circles) and K299Q (orange filled triangles).