To overview τ-N-histidine methylation in cells and the corresponding enzymes, we quantified τ-N-methylhistidine by mass spectrometry (MS). To distinguish the two types of histidine methylations, we performed multiple reaction monitoring (MRM) of digested amino acids to trace specific m/z transitions from the precursor (Davydova et al., 2021) (see ‘Materials and methods’ section for details). Because SETD3 modifies abundant actin proteins (Dai et al., 2019; Guo et al., 2019; Kwiatkowski et al., 2018; Wilkinson et al., 2019; Zheng et al., 2020), SETD3 knockout (KO) in HEK293T cells (Figure 1—figure supplement 1B–E) greatly reduced τ-N-methylhistidine (Figure 1A). However, a substantial fraction of τ-N-methylhistidine was left in the SETD3 KO cells (Figure 1A), suggesting the presence of other mammalian τ-N-methyltransferase(s).

An apparent candidate for the remaining τ-N-methyltransferase in SETD3 KO cells was METTL18, whose yeast homolog (histidine protein methyltransferase 1 [Hpm1]) has been reported to catalyze τ-N-methylation on histidine in Rpl3 (Al-Hadid et al., 2016; Al-Hadid et al., 2014; Webb et al., 2010). Indeed, the SETD3-METTL18 double-KO (DKO) cells (Figure 1—figure supplement 1B–E) had even lower levels of τ-N- methylhistidine than the SETD3 single-KO cells (Figure 1A). Note that we could not detect any significant alteration in π-N-methylhistidine in any cells tested in this study (Figure 1—figure supplement 1F).

These observations led us to survey the methylation substrate of METTL18. For this purpose, we harnessed propargylic Se-adenosyl-l-selenomethionine (ProSeAM), an analog of S-adenosyl-l-methionine (SAM). This compound acts as a donor in the methylation reaction by the SET domain and 7-β-strand methyltransferase (Davydova et al., 2021; Shimazu et al., 2014; Shimazu et al., 2018). Instead of a methyl moiety, a propargyl unit was added to the substrate residue, allowing biotin tagging with a click reaction (Davydova et al., 2021; Shimazu et al., 2014; Shimazu et al., 2018; Figure 1B). Using ProSeAM, we performed in vitro methylation with recombinant METTL18 (Figure 1—figure supplement 1G) in the lysate of METTL18 KO cells and detected the ProSeAM-reacted proteins in a METTL18 protein-dependent manner (Figure 1C). The high background signals are likely to have originated from methyltransferases other than METTL18 in the lysate.

For the quantitative and sensitive proteomic identification of methylated protein(s), we combined this approach with stable isotope labeling using amino acids in cell culture (SILAC) (Figure 1B). ProSeAM was added to the cell lysates of METTL18 KO cells labeled with either light or heavy isotopic amino acids. Only the extract from heavy-isotope-labeled cells was incubated with recombinant METTL18 (Figure 1B). After lysate pooling and streptavidin purification (Figure 1B), isolated proteins were quantitatively assessed by liquid chromatography (LC)-MS/MS. Among the detected proteins, human RPL3 was the only candidate to be reproducibly identified (Figure 1D).

To ensure cellular histidine methylation on the protein, FLAG-tagged RPL3 was expressed in naïve HEK293T cells, immunopurified, and subjected to LC-MS/MS. This analysis identified the methylhistidine in cellular RPL3 and precisely annotated the residue at His245 (Figure 1E). In stark contrast, the same experiments with METTL18 KO did not detect methylated His245 (Figure 1F). The ectopic expression of wild-type (WT) METTL18 in the KO cells rescued the modification, whereas mutations in the potential SAM binding site (Asp193Lys-Gly195Arg-Gly197Arg), which were predicted based on lysine methyltransferase orthologs (Ng et al., 2002), abolished the potential (Figure 1F). The same METTL18-dependent His245 methylation was also found in endogenous RPL3 (Figure 1—figure supplement 2C and D), isolated in the 60S subunit (Figure 1—figure supplement 2A and B).

To distinguish the methylation forms on histidine, we applied MRM to the peptide fragments containing His245 and clearly observed that methylation occurs at the τ-N position but not at the π-N position (Figure 1G).

Thus, taken together, our data demonstrated that His245 of RPL3 is a METTL18 substrate for the formation of τ-N-methylhistidine in cells.

As RPL3 in crude lysate was methylated in vitro (Figure 1C and D), we set out to recapitulate this reaction by purified factors using ¹⁴C-labeled SAM as a methyl donor. Irrespective of the human or mouse homolog, the immunopurified FLAG-tagged RPL3 proteins transiently expressed in METTL18 KO cells were efficiently labeled by recombinant METTL18 (Figure 2A). Moreover, changing His245 to Ala completely abolished the reaction, validating His245 as a methylation site.

In contrast, recombinant RPL3 expressed in bacteria was a poor substrate (Figure 2A). A similar deficiency in the in vitro methylation assay by Hpm1 on solo RPL3 protein produced in bacteria was seen in yeast (Al-Hadid et al., 2014). Given that RPL3 expressed in mammalian cells assembled into ribosomes but RPL3 expressed in bacteria did not, we hypothesized that METTL18 recognizes RPL3 within ribosomes or preassembled intermediates. To characterize the molecular complex associated with METTL18, we separated the ribosomal complexes through a sucrose density gradient and found that METTL18 was associated with a complex smaller than mature 60S (Figure 2B). Given the smaller size, we speculated that the METTL18-associating complex is pre-60S in the middle of biogenesis. Indeed, the METTL18-containing subfractions also possessed PES1 (a yeast Nop7 homolog), a ribosome biogenesis factor at an early step (Kater et al., 2017; Sanghai et al., 2018), and a small portion of RPL3 (Figure 2B). On the other hand, NMD3, an adaptor protein of the late-stage pre-60S for cytoplasmic export (Ma et al., 2017; Malyutin et al., 2017), was almost exclusive to the METTL18-associating pre-60S. These data suggested that RPL3, in an early intermediate complex of 60S biogenesis, is an efficient substrate for METTL18.

The presence of METTL18 in the pre-60S complex led us to investigate the role of methylation in ribosome biogenesis. However, assessed by the bulk 28S rRNA abundance (Figure 3—figure supplement 1A) and the 60S fraction in sucrose density gradient (Figure 3—figure supplement 1B and C), no altered abundance of the 60S subunit (relative to the 40S subunit) was observed in METTL18 KO cells, while the knockdown of RPL17, which is known to hamper 60S biogenesis (Wang et al., 2015), reduced the 60S abundance in a sucrose density gradient (Figure 3—figure supplement 1D–F).

Thus, we hypothesized that RPL3 methylation may impact protein synthesis. To understand the potential role of methylation in RPL3, we reanalyzed published cryo-EM data (Osterman et al., 2020) and assessed the τ-N-methyl moiety on His245 of RPL3 (Figure 3A), which indicated that τ-N-methylated RPL3 is a stoichiometric component of the ribosome. A similar density of methylation on the histidine of RPL3 has also been found in rabbit ribosomes (Bhatt et al., 2021). On the other hand, the ribosomes isolated from METTL18 KO cells lost density at the τ-N position of the His245 (Figure 3B, Figure 3—figure supplement 2, Table 1). The absence of methylation at the τ-N position allowed nitrogen to form hydrogen bonds with G1595 of 28S rRNA (Figure 3B). Given that G1595 is located in the loop of helix 35, which macrolide antibiotics target in bacterial systems (Kannan and Mankin, 2011), this difference in the interaction between His245 and G1595 suggests an alteration in the translation reaction.

To investigate the impacts of the modification on protein synthesis, we assessed global translation in METTL18 KO cells. However, we could not detect a significant difference in overall translation probed by polysome formation (Figure 4—figure supplement 1A and B), whereas RPL17 knockdown reduced the polysome (Figure 4—figure supplement 1C and D). Similarly, we could not observe a significant alteration in nascent peptides labeled with O-propargyl-puromycin (OP-puro) (Figure 4—figure supplement 1E).

Nonetheless, we investigated the implications of RPL3 methylation defects across the transcriptome by ribosome profiling (Ingolia et al., 2009; Iwasaki and Ingolia, 2017). Strikingly, we found increased translation elongation of Tyr codons in METTL18 KO cells; ribosome occupancy on Tyr codons at the A-site was selectively reduced in the mutant cells (Figure 4A and B). This trend in ribosome occupancy was not observed at the P and E sites (Figure 4—figure supplement 2A and B). To evaluate the amino acid context associated with the high elongation rate, we surveyed the motifs around the A-site with reduced ribosome occupancy by METTL18 depletion (Figure 4C) and analyzed the enriched/depleted sequence in the group (Figure 4D). Remarkably, Tyr at the A-site was the predominant determinant for fast elongation in METTL18 KO cells (Figure 4D).

The smoother elongation may be caused by increased tRNA abundance. However, tRNA^Tyr_GUA, which decodes both UAU and UAC codons, was noticeably reduced by METTL18 depletion (Figure 4—figure supplement 2C and D). On the other hand, the abundance control tRNA^Leu_HAG was not altered (Figure 4—figure supplement 2E and F). Thus, tRNA abundance did not explain the reduced footprints on Tyr codons.

Further ribosome profiling data analysis supported that the accommodation of tRNA at the A-site may not explain the modulation of elongation by RPL3 methylation. Ribosome footprints possess two distinct populations of different lengths (short, peaked at ~22 nt; long, peaked at ~29 nt), reflecting the presence of A-site tRNA (Lareau et al., 2014; Wu et al., 2019; Figure 4E). Since A-site tRNA-free ribosomes are more susceptible to RNase treatment at the 3′ end, the trimmed short footprints represent nonrotated ribosomes waiting for A-site codon decoding by tRNA. On the other hand, long footprints originate from ribosomes accommodated with A-site tRNA in the middle of the peptidyl transfer reaction or subsequent subunit rotation (Lareau et al., 2014; Wu et al., 2019). The reduced footprints on Tyr codons in METTL18 KO cells were more prominent in long footprints than short footprints (Figure 4F). Moreover, the long footprints were recovered by ectopic expression of METTL18 protein (Figure 4G). These data suggested that the process or processes downstream of codon decoding are promoted in KO cells. This observation was reasonable, given the relatively long distance between the τ-N-methylated histidine and the mRNA codon.

To test the cell-type specificity, we further extended our ribosome profiling experiments in HAP1 cells, which originate from chronic myelogenous leukemia. We investigated ribosome footprints on the Tyr codons in three independent METTL18 KO cell lines (1-nt, 2-nt, and 4-nt deletions) that should lead to truncated proteins (Figure 4—figure supplement 2G) and observed the reduction of ribosome occupancies on the codons in those cell lines as well (Figure 4H). These data suggested consistent effects of RPL3 methylation on elongation irrespective of cell type.

To further assess the effect of RPL3 methylation on Tyr codon elongation, we combined hybrid in vitro translation (Erales et al., 2017; Panthu et al., 2015) with the Renilla-firefly luciferase fusion reporter system (Kisly et al., 2021; Figure 5A). In this setup, we purified ribosomes from HEK293T (naïve or METTL18 KO) cells, added them to ribosome-depleted rabbit reticulocyte lysate (RRL), and then conducted an in vitro translation assay (i.e., hybrid translation) (Erales et al., 2017; Panthu et al., 2015; Figure 5A). Indeed, we observed that removal of the ribosomes from RRL decreased protein synthesis in vitro and that complementation with ribosomes from HEK293T cells recovered the activity (Figure 5—figure supplement 1A).

To test the effect on Tyr codon elongation, we harnessed the fusion of Renilla and firefly luciferases; this system makes it possible to detect the delay/promotion of downstream firefly luciferase synthesis compared to upstream Renilla luciferase and thus to focus on the effect of the sequence inserted between the two luciferases on elongation (Kisly et al., 2021; Figure 5A). For better detection of the effects on Tyr codons, we used a repeat of the codon (×39; the number was due to cloning constraints). We note that the insertion of Tyr codon repeats reduced the elongation rate (or processivity), as we observed the reduced slope of the downstream Fluc synthesis (Figure 5—figure supplement 1B).

Then, we coupled the in vitro translation system and the reporter. We observed that RPL3 methylation-deficient ribosomes exhibited faster elongation on Tyr repeats than ones from naïve cells (Figure 5B).

These in cell and in vitro data together indicated that RPL3 methylation mediated a slowing of translation elongation on the Tyr codons.

Ribosome traverse along the mRNA determines the quality of protein synthesized (Cassaignau et al., 2020; Collart and Weiss, 2019; Stein and Frydman, 2019). Slowdown of ribosome elongation is advantageous since it allows the duration of nascent protein folding before completion of protein synthesis. Therefore, we reasoned that translation elongation slowdown at Tyr codons by RPL3 methylation facilitates protein folding on ribosomes and maintains proper homeostasis of the proteome.

To test this possibility, we employed an aggregation-prone firefly luciferase (Fluc) reporter (Arg188Gln-Arg261Gln double mutant or DM) fused to enhanced green fluorescent protein (EGFP) (Gupta et al., 2011). Since this engineered protein necessitates chaperones to fold properly, the reduced pool of available chaperones (i.e., proteotoxicity) leads to aggregation puncta of the reporter protein in cells. Indeed, METTL18 depletion induced the aggregation of FlucDM but not Fluc WT reporter (Figure 6A and B). As shown by the recovery of protein abundance by treatment with the proteasome inhibitor MG132 (Figure 6C), FlucDM protein was synthesized in METTL18 KO cells of low quality and subjected to degradation.

Then, we explored the proteome, the quality of which was assisted by RPL3 histidine methylation. For this purpose, we surveyed the proteins aggregated in METTL18 KO cells by SILAC (Figure 7A). We observed that a subset of proteins were enriched in precipitates of METTL18 KO cell lysates (Figure 7B). This subgroup significantly accumulated Tyr-rich proteins (defined as proteins possessing 30 Tyr or more) (hypergeometric test, p=0.0068) (Figure 7B). The high probability of protein precipitates could not be explained by the increased net protein synthesis measured by ribosome profiling (Figure 7—figure supplement 1A). More generally, Tyr-rich proteins were more prone to precipitate upon the deletion of METTL18 (Figure 7—figure supplement 1B) than proteins rich in other amino acids (Figure 7C). Thus, proteomic analysis of cellular precipitates revealed that the modulation of Tyr-specific translation elongation by RPL3 methylation confers proteome integrity.

These data led us to further investigate the properties of the proteome associated with Tyr. Here, we surveyed the aggregation propensity of the precipitated proteins by TANGO, which is based on statistical mechanics (Fernandez-Escamilla et al., 2004). Strikingly, metagene analysis showed prominent enrichment of the TANGO-predicted aggregation propensity around the Tyr codons (Figure 7D). As exemplified in the MACROH2A1 protein, the subpart of the aggregation-prone region in this protein was found on Tyr with reduced ribosome occupancies in METTL18 KO cells (Figure 7E, Figure 7—figure supplement 1C and D).

Ultimately, the low-quality protein synthesized by RPL3-unmethylated ribosomes may serve as a substrate for proteasomes. To assess the fraction of proteins degraded by proteasome, we conducted SILAC analysis of the total proteome in METTL18 KO cells treated with MG132, a proteasome inhibitor (Figure 8A). Indeed, this approach revealed a subset of more efficiently degraded proteins in METTL18 KO cells (Figure 8B); this group of proteins tended to be reduced in METTL18 KO cells and recovered by MG132 treatment. Moreover, this subset of proteins included Tyr-rich proteins (Figure 8C), which was congruent with the enhanced elongation observed in ribosome profiling.

The relationships among the aggregation-prone character of the protein, the cellular protein precipitates, increased proteasomal substrates, and reduced ribosome occupancy indicates that Tyr translation modulated by RPL3-methylated ribosomes is associated with the quality of the synthesized proteins, preventing unwanted protein aggregation (Figure 9).

HEK293T KO cell lines were generated by the CRISPR-Cas9 system. All cell lines in the laboratory were routinely tested for Mycoplasma contamination with the MycoAlert Mycoplasma Detection Kit (LONZA) and confirmed negative.

Anti-α-tubulin (Sigma-Aldrich, clone B-5-1-2, 1:1000), anti-METTL18 (Proteintech Group, 25553-1-AP), anti-SETD3 (Abcam, ab174662, 1:1000), anti-RPL3 (Proteintech Group, 66130-1-lg and 11005-1-AP, 1:1000), anti-PES1 (Abcam, ab252849, 1:1000), anti-NMD3 (Abcam, ab170898, 1:1000), anti-HA (Medical & Biological Laboratories [MBL], M180-3, 1:1000), anti-GFP (Abcam, ab6556, 1:1000), anti-RPL17 (Proteintech Group, 14121–1-AP, 1:1000), and anti-β-actin (MBL, M177-3, 1:1000) primary antibodies were used.

For Figures 1C and 2A, Figure 1—figure supplement 1D, anti-mouse IgG, HRP-Linked Whole Ab Sheep (Cytiva, NA931V, 1:5000) and anti-rabbit IgG, HRP-Linked Whole Ab Donkey (Cytiva, NA934V, 1:5000) secondary antibodies were used. The chemiluminescence was raised with a Western Lightning Plus-ECL Kit (PerkinElmer) according to the manufacturer’s protocol and detected with X-ray film (FUJIFILM, RX-U). For biotinylated proteins, high-sensitivity streptavidin-HRP (Thermo Fisher Scientific, 21130) was used.

To generate Figures 2B and 5C, Figure 3—figure supplement 1D, IRDye680- or IRDye800CW-conjugated secondary antibodies (LI-COR Biosciences, 925-68070/71 and 926-32210/11/19, respectively, 1:10,000) were used. Images were obtained with Odyssey CLx (LI-COR Biosciences).

ProSeAM was synthesized as previously described (Sohtome et al., 2018). ProSeAM substrate screening was carried out as reported (Shimazu et al., 2018) with some modifications. METTL18 KO cells were cultured in Dulbecco’s modified Eagle’s medium (DMEM) containing either light Arg/Lys or heavy isotope-labeled Arg (¹³C₆ ¹⁵N₄ l-arginine)/Lys (¹³C₆ ¹⁵N₂ l-lysine) (Thermo Fisher Scientific, respectively) at least six doubling times. The cells were lysed with 50 mM Tris-HCl pH 8.0, 50 mM KCl, 10% glycerol, and 1% n-dodecyl-β-d-maltoside. For the cells cultured with the heavy amino acids, the lysate containing 200 μg of proteins was incubated with 150 μM ProSeAM and 10 μg of His-METTL18 in 50 mM Tris-HCl pH 8.0 at 20°C for 2 hr. For the lysate with the light amino acids, His-METTL18 was omitted from the reaction. The reaction was stopped by the addition of four volumes of ice-cold acetone. The proteins were precipitated by centrifugation at 15,000 × g for 5 min, washed once with ice-cold acetone, and then dissolved in 58.5 µl of 1× PBS containing 0.2% sodium dodecyl sulfate (SDS). After the addition of 15 µl of 5× click reaction buffer (7.5 mM sodium ascorbate [Nacalai Tesque], 0.5 mM TBTA [AnaSpec], and 5 mM CuSO₄) and 1.5 µl of 10 mM Azide-PEG4-Biotin (Click Chemistry Tools), the click reaction was conducted for 60 min at room temperature, stopped with four volumes of ice-cold acetone, and precipitated as described above. The protein in the pellet was resuspended in 75 µl of binding buffer (1× PBS, 0.1% Tween-20, 2% SDS, and 20 mM DTT) and sonicated for 10 s. The light and heavy isotope-labeled samples were pooled in 450 µl of IP buffer (Tris-buffered saline [TBS] and 0.1% Tween-20) containing 3 µg of Dynabeads M-280 Streptavidin (Thermo Fisher Scientific) and incubated for 30 min at room temperature (note that the final SDS concentration in the solution was 0.5%). The protein-bound beads were washed three times with bead wash buffer (1× PBS, 0.1% Tween-20, and 0.5% SDS) and twice with 100 mM ammonium bicarbonate (ABC) and then used for Western blotting and MS.

For MS/MS analysis, the beads were incubated in 20 mM DTT and 100 mM ABC for 30 min at 56°C and then for 30 min at 37°C in the dark with supplementation with 30 mM iodoacetamide. Subsequently, proteins were digested with 1 µg trypsin (Promega) and subjected to LC (Thermo Fisher Scientific, EASY-nLC 1000) coupled to a Q Exactive Hybrid Quadrupole-Orbitrap Mass Spectrometer (Thermo Fisher Scientific) with a nanospray ion source in positive mode, as previously reported (Davydova et al., 2021). The peptides loaded on a NANO-HPLC C18 capillary column (0.075 mm inner diameter × 150 mm length, 3 µm particle size, Nikkyo Technos) were eluted at a flow rate of 300 nl/min with the two different slopes of mobile phase ‘A’ (water with 0.1% formic acid) and mobile phase ‘B’ (acetonitrile with 0.1% formic acid): 0–30% of phase B in 100 min and 30–65% of phase B in 20 min. Subsequently, the mass spectrometer in the top 10 data-dependent scan mode was run with the following parameters: spray voltage, 2.3 kV; capillary temperature, 275°C; mass-to-charge ratio, 350–1800; normalized collision energy, 28%. The MS and MS/MS data obtained with Xcalibur software (Thermo Fisher Scientific) were surveyed in the Swiss-Prot database with Proteome Discoverer (version 2.3, Thermo Fisher Scientific) and MASCOT search engine software (version 2.7, Matrix Science). Peptides with false discovery rates (FDRs) less than 1% were considered. Proteins with the following criteria were defined as METTL18-dependent labeled proteins: 1.5-fold or more increase in heavy amino acid sample compared to light amino acid sample; one, 10% or more coverage of the protein; and three or more peptides identified.

The SDS-PAGE-separated and Coomassie staining-visualized proteins were excised and destained. The gel slices were reduced with 50 mM DTT and 4 M guanidine-HCl at 37°C for 2 hr, followed by alkylation with 100 mM acrylamide at 25°C for 30 min. The RPL3 proteins were digested with chymotrypsin. The MS and MS/MS spectra were acquired with a Q Exactive HFX (Thermo Fisher Scientific). The mass spectrometer was operated in positive mode. The MS/MS spectra were obtained using a data-dependent top 10 method. The acquired data were processed using Proteome Discoverer (version 2.3, Thermo Fisher Scientific). The processed data were used to search with MASCOT (version 2.7, Matrix Science) against the in-house database including the amino acid sequences of RPL3 using the following parameters: type of search, MS/MS ion search; enzyme, none; fixed modification, none; variable modifications, Gln->pyro-Glu (N-term Q), oxidation (M), oropionamide (C), and methyl (H); mass values, monoisotopic; peptide mass tolerance, ±15 ppm; fragment mass tolerance, ±30 mmu; peptide charge, 1+, 2+, and 3+; instrument type, ESI-TRAP.

Isotopically heavy amino acids (0.1  mg/ml ¹³C₆¹⁵N₂ l-lysine-HCl and 0.1  mg/ml ¹³C₆¹⁵N₄ l-arginine-HCl [both FUJIFILM Wako Chemicals]) or regular amino acids (0.1  mg/ml l-lysine-HCl and 0.1  mg/ml l-arginine-HCl [both FUJIFILM Wako Chemicals]) were added to DMEM deficient in both l-lysine and l-arginine for SILAC (Thermo Fisher Scientific) supplemented with 10% dialyzed fetal bovine serum (FBS) (Sigma-Aldrich). Naïve HEK293T cells and METTL18 KO cells were cultured in media with heavy and light isotopes, respectively, for 2 weeks.

After a brief wash with PBS, cells were lysed with buffer containing 20 mM Tris-Cl pH 7.5, 150 mM KCl, 5 mM MgCl₂, 1% Triton X-100, and 1 mM DTT. The protein concentration was measured using Qubit 2.0 Fluorometer (Thermo Fisher Scientific). The same amounts of proteins from light and heavy isotope-labeled lysates were mixed and centrifuged at 500 × g for 3 min. The supernatant was further centrifuged at 20,000 × g and 4°C for 15 min. The precipitate was collected and used for analysis.

Filter-aided sample preparation (FASP) was used for protein digestion (Wiśniewski et al., 2009). The precipitate was dissolved in SDT lysis buffer (4% [w/v] SDS, 0.1 M DTT, and 100 mM Tris-HCl pH 7.6) and incubated at 95°C for 5 min. Denatured samples were diluted 10 times with UA buffer (8 M urea and 100 mM Tris-HCl pH 8.5) and loaded to Vivacon 500, 30000 MWCO Hydrosart (Sartorius) to trap the protein on the filter. The filter unit was washed with the UA buffer. Proteins on the filter unit were alkylated with 100 μl of the IAA solution (0.05 M iodoacetamide in UA). Then, the filter unit was washed with 100 μl of UA buffer three times and 100 μl of 50 mM NH₄HCO₃ three times. Proteins on the filter unit were digested with 40 μl of trypsin solution (trypsin [V5111, Promega] in 50 mM NH₄HCO₃) at 37°C for 18 hr. Then, the peptides were eluted from the filter unit by 40 μl of NH₄HCO₃ twice and 50 μl of 0.5 M NaCl once and then pooled.

LC-MS/MS analysis was performed using EASY-nLC 1000 (Thermo Fisher Scientific) and Q Exactive (Thermo Fisher Scientific) equipped with a nanospray ion source. The peptides were separated with a NANO-HPLC capillary column C18 (0.075 × 150 mm, 3 μm, Nikkyo Technos) at 300 nl/min flow rate with strep gradients of solvent A (0.1% formic acid) and solvent B (acetonitrile with 0.1% formic acid); 0–30% B for 100 min and then 30–65% B for 20 min. The resulting MS and MS/MS data were searched against the Swiss-Prot database using Proteome Discoverer (version 2.4, Thermo Fisher Scientific) with MASCOT search engine software (version 2.7, Matrix Science). The peptides with FDR values of 0.05 or less were considered for the subsequent analysis. To quantify the difference of peptides labeled with differential isotopes, the built-in SILAC 2-plex quantification method in Proteome Discoverer (version 2.4, Thermo Fisher Scientific) was used. Peptide abundances were normalized by total peptide amount. p-Values were calculated by the background-based t-test and then adjusted by the Benjamini–Hochberg method.

For total proteome analysis, we treated cells with 0.25 μM MG132 (FUJIFILM Wako Chemicals) for 24 hr before cell harvest. Cell lysates were prepared as described above, and the supernatant obtained by centrifugation at 20,000 × g and 4°C for 15 min was used for the downstream analysis.

FLAG-tagged RPL3 was transiently expressed in METTL18 KO cells and immunopurified with ANTI-FLAG M2 Affinity Gel (Sigma-Aldrich). The purified proteins were incubated in 1× reaction buffer (50 mM Tris-HCl pH 8.5 and 50 mM MgCl₂) with 1 µg of His-GST-METTL18, 2 µM MTAN, and 0.01 µCi of ¹⁴C-labeled SAM (PerkinElmer) at 30°C for 2 hr. The reaction was stopped by the addition of Laemmli SDS-sample buffer. Proteins were separated on a 10% acrylamide SDS-PAGE gel. The dried gel was exposed to an imaging plate (FUJIFILM) for 48 hr. The autoradiograph was detected with a phosphor imaging scanner (Cytiva, Amersham Typhoon).

The crude ribosomal pellet was suspended in buffer A (50 mM Tris-HCl pH 7.5, 150 mM KCl, 4 mM magnesium acetate, 1 mM DTT, 7% [w/v] sucrose, and 1 mM puromycin) by mixing with a magnetic stirrer for 4 hr on ice. After stirring, the suspension was centrifuged at 15,000 × g for 10 min at 4°C. The supernatant was loaded onto a HiPrep 16/60 Sephacryl S-500 column (Cytiva) equilibrated with buffer A without puromycin. The fraction was recovered and concentrated using Amicon Ultra (MWCO 50 kDa, Merck Millipore). The concentrated mixture was supplemented with 0.1 mM puromycin, 2 mM magnesium acetate, and 2 mM ATP and incubated for 30 min at 37°C. The reaction was loaded onto a 10–40% (w/v) sucrose gradient with buffer B (50 mM Tris-HCl pH 7.5, 500 mM KCl, 4 mM magnesium acetate, and 2 mM DTT) and centrifuged at 25,000 rpm in an SW28 rotor for 16 hr at 4°C. 2 ml fractions were successively fractionated from the top of the gradient. An aliquot of each fraction was analyzed by SDS-PAGE, and the gels were stained with Coomassie brilliant blue (CBB) to detect the 40S and 60S subunit proteins. The 40S- and 60S-subunit fractions were mixed and concentrated using Amicon Ultra filter units (MWCO 50 kDa, Merck Millipore). The concentrated mixture was loaded onto a 10–50% (w/v) sucrose gradient with buffer C (50 mM Tris-HCl pH 7.5, 150 mM KCl, 10 mM magnesium acetate, and 2 mM DTT) and centrifuged at 28,000 rpm in an SW41Ti rotor for 3 hr at 4°C. Gradient fractionation was carried out using a piston gradient fractionator equipped with a TRIAX flow cell detector (BioComp) by continuous monitoring of absorbance at a wavelength of 280 nm. The fractions containing 80S ribosomes were dialyzed against preparation buffer (10 mM HEPES-KOH pH 7.5, 30 mM potassium acetate, 10 mM magnesium acetate, and 1 mM DTT) and concentrated using Amicon Ultra filter units (MWCO 50 kDa, Merck Millipore). The concentrated sample was quantified by measuring absorbance at a wavelength of 260 nm and flash-cooled with liquid nitrogen.

For cryo-EM sample preparation, Quantifoil R1.2/1.3 300 mesh copper grids (Quantifoil) were covered with an amorphous carbon layer prepared in-house. The thawed sample was diluted to 70 nM (an absorbance at 260 nm of 3.5) with preparation buffer, and 3 μl of the sample was applied onto grids at 4°C at 100% relative humidity using a Vitrobot Mark IV (FEI). After incubation for 30 s and blotting for 3 s, the grids were plunged into liquid ethane.

The cryo-EM dataset was collected with a Tecnai Arctica transmission electron microscope (FEI) operated at 200 kV using a K2 summit direct electron detector (Gatan) (0.97 Å/pixel). The 5517 images collected were fractionated to 40 frames, with a total dose of ~50 e^-/Ų.

Processing of cryo-EM data was performed with RELION-3.1 (Zivanov et al., 2020). The movie frames were aligned with MotionCor2 (RELION’s own implementation), and the CTF parameters were estimated with CTFFIND-4.1 (Rohou and Grigorieff, 2015). Particles were automatically picked using a template-free Laplacian-of-Gaussian (LoG) filter (250–500 Å), and 381,227 particles were extracted with twofold binning. After 2D classification, 136,688 particles were selected and applied to 3D classification. A low-pass-filtered (40 Å) map of the human 80S ribosome (EMD-9701) (Yokoyama et al., 2019) was used as a reference map for 3D classification, and 118,470 particles were selected after this step. These particles were re-extracted without rescaling and used in 3D refinement, Bayesian polishing, CTF refinement, and then 3D refinement again. After these steps, focused refinement with a mask on the 60S subunit and postprocessing resulted in a resolution of 2.72  Å.

For molecular modeling, the model of the 60S subunit from the human 80S ribosome structure at 2.9 Å resolution (PDB: 6QZP) (Natchiar et al., 2017) was used as a starting model and manually fitted into the map using UCSF Chimera (Pettersen et al., 2004). Map sharpening and model refinement were performed in PHENIX (Adams et al., 2010), and the model was further refined manually with Coot (Emsley et al., 2010). The modified nucleotides of ribosomal RNAs were introduced based on quantitative MS data (Taoka et al., 2018).

For Figure 2B, cells were lysed with whole-cell lysis buffer (50 mM Tris-HCl pH 7.5, 150 mM NaCl, 5 mM MgCl₂, 1% NP-40, 1 mM DTT, and 100 µg/ml cycloheximide). The whole-cell lysates were passed through a 21-gauge needle and incubated at 4°C for 15 min. To obtain Figure 4—figure supplement 1A–D and Figure 9—figure supplement 2A and B, cells were lysed with lysis buffer (20 mM Tris-Cl pH 7.5, 150 mM KCl, 5 mM MgCl₂, 1% Triton X-100, 1 mM DTT, and 100 µg/ml cycloheximide) and cleared by centrifugation at 20,000 × g and 4°C for 10 min. Cell lysate containing 40 µg of total RNA was loaded onto a 10–50% sucrose gradient and ultracentrifuged at 35,300 rpm and 4°C for 2.5 hr by a Himac CP80WX ultracentrifuge (Hitachi) with a P40ST rotor (Hitachi).

To obtain Figure 1—figure supplement 2A, Figure 3—figure supplement 1B, C, E, and F, and Figure 9—figure supplement 2C and D, cells were lysed with EDTA lysis buffer (20 mM Tris-Cl pH 7.5, 150 mM KCl, 5 mM EDTA, 1% Triton X-100, and 1 mM DTT) and centrifuged at 20,000 × g and 4°C for 10 min. The supernatant with 40 µg of total RNA was ultracentrifuged at 35,300 rpm and 4°C for 2.5 hr (Figure 1—figure supplement 2A) or at 38,000 rpm and 4°C for 4 hr (Figure 3—figure supplement 1B, C, E, and F and Figure 9—figure supplement 2C and D) by a Himac CP80WX ultracentrifuge (Hitachi) with a P40ST rotor (Hitachi).

To knock down RPL17, HEK293T cells were seeded in complete growth medium in 10 cm² dishes 24 hr before siRNA transfection. Then, the cells were transfected with RPL17 siRNA (ON-TARGETplus Human RPL17 siRNA, L-013633-01-0005, Horizon Discovery) or control siRNA (ON-TARGETplus Non-targeting Control Pool, D-001810-10-50, Horizon Discovery) by TransIT-X2 Dynamic Delivery System (Mirus) according to the manufacturer’s instructions. After 72 hr of siRNA transfection, cells were lysed by the procedure described above.

The gradients were fractionated with continuous measurement of the absorbance at 260 nm by a Triax Flow Cell (BioComp) and Micro collector (ATTO). For ribosomal complex quantification, the straight line connecting the valleys before and after peaks was defined as a background baseline. Then, the area enclosed by the baseline and the peak waveform was calculated. For Western blot (Figure 2B) or MS analysis (Figure 1—figure supplement 2B), fractions were concentrated using Amicon Ultra filter units (MWCO 10 kDa, Millipore).

Metabolic labeling of nascent proteins with OP-puro was performed as previously described (Iwasaki et al., 2019). Cells were treated with 20 µM OP-puro and incubated at 37°C for 30 min in a CO₂ incubator. After washing with PBS, cells were lysed with buffer containing 20 mM Tris-HCl pH 7.5, 150 mM NaCl, 5 mM MgCl₂, and 1% Triton X-100. Nascent polypeptides were labeled with IRdye800CW Azide (LI-COR Biosciences) with a Click-iT Cell Reaction Buffer Kit (Thermo Fisher Scientific). After free dye was removed by a G-25 column (Cytiva), labeled polypeptides were separated by SDS-PAGE. The gel was imaged by Odyssey CLx (LI-COR Biosciences) for the detection of nascent peptides with infrared at 800 nm. Then, total proteins were stained with CBB (FUJIFILM Wako Chemicals) and imaged with an infrared 700 nm signal. The gel area ranging from 17 kDa to 280 kDa was quantified using Image Studio (version 5.2, LI-COR Biosciences), and the nascent peptide signal was normalized to the total protein signal.

Total RNA was extracted from cells by TRIzol Reagent (Thermo Fisher Scientific) according to the manufacturer’s instructions. Purified RNAs were electrophoresed on Super Sep RNA gels (FUJIFILM Wako Chemicals), transferred onto nylon membranes (Biodyne, Thermo Fisher Scientific), and then UV-crosslinked. The membranes were incubated with UltraHyb-Oligo (Thermo Fisher Scientific) at 37°C for 1 hr. DNA oligonucleotide probes (see below for details) were radiolabeled with [γ−³²P] ATP (PerkinElmer) by T4 PNK (New England Biolabs) and purified with a G-25 column (Cytiva). After prehybridization, membranes were incubated with a labeled DNA probe overnight at 37°C and washed three times with 2× saline-sodium citrate solution. The signal on the membranes was detected by an Amersham Typhoon (Cytiva) scanner. DNA oligonucleotide sequences used as probes are listed below:

tRNA^Tyr_GUA: 5′-ACAGTCCTCCGCTCTACCAGCTGA-3′, tRNA^Leu_HAG: 5′-CAGCGCCTTAGACCGCTCGGCCA-3′, and U6: 5′-CACGAATTTGCGTGTCATCCTT-3′.

The hybrid translation assay was performed as described previously (Erales et al., 2017; Panthu et al., 2015) with optimization for the purpose of this study.

For Fluc reporter, a PCR fragment was PCR-amplified from pGL3 basic (Promega) with the primers 5′-TGACTAATACGACTCACTATAGG-3′ and 5′-TGTATCTTATCATGTCTGCTCGAA-3′. For Rluc-Y0×-Fluc and Rluc-Y39×-Fluc reporters, DNA fragments were PCR-amplified by psiCHECK2-Y0× and Y39× with the primers 5′-TGACTAATACGACTCACTATAGG-3′ and 5′-TGTATCTTATCATGTCTGCTCGAA-3′.

The DNA fragments were used for in vitro transcription by T7-Scribe Standard RNA IVT kit (CELLSCRIPT). Then, the RNAs were capped with the ScriptCap m⁷G Capping system (CELLSCRIPT) and polyadenylated with the A-Plus poly(A) polymerase Tailing Kit (CELLSCRIPT).

1 ml of RRL (nuclease-treated, Promega) was ultracentrifuged at 240,000 × g for 2 hr 15 min at 4°C by an Optima MAX-TL ultracentrifuge (Beckman Coulter) with a TLA110 rotor (Beckman Coulter). Then, 900 μl of supernatant was collected, flash-frozen by liquid nitrogen, and stored at −80°C.

After brief washing with PBS, cells were resuspended with PBS and centrifuged at 500 × g for 3 min. The pellet of cells was resuspended in the same volume of buffer R (10 mM HEPES pH 7.5, 10 mM KOAc, 1 mM MgOAc₂, and 1 mM DTT) and incubated for 15 min on ice. Subsequently, the mixture was vortexed for 30 s and centrifuged at 16,000 × g for 10 min at 4°C. Then, 300 μl of the supernatant was ultracentrifuged through a 1 ml sucrose cushion (1 M sucrose in buffer R) at 240,000 × g for 2 hr 15 min at 4°C by an Optima MAX-TL ultracentrifuge (Beckman Coulter) with a TLA110 rotor (Beckman Coulter). The pellet was rinsed with buffer R2 (20 mM HEPES pH 7.5, 10 mM NaCl, 25 mM KCl, 1.1 mM MgCl₂, and 7 mM 2-mercaptoethanol) and then resuspended in 30 μl of buffer R2 by stirring at 4°C for 15 min, flash-frozen by liquid nitrogen, and stored at −80°C. The ribosome concentration was determined by the absorbance at 260 nm using a NanoDrop (Thermo Fisher Scientific).

In vitro translation was typically performed in a 10 μl reaction mixture containing 5 μl of RRL supernatant, 22.6 nM of purified ribosome, 1.1 nM of reporter mRNA, 75 mM KCl, 0.75 mM MgCl₂, and 20 µM of amino acid mixture (Promega).

For the Fluc reporter, the reaction mixture was incubated at room temperature for 90 min and stopped by the addition of 20 μl of 1× Passive Lysis Buffer (Promega). The FLuc luminescence from 30 μl of the mixture was measured with the Dual-Luciferase Reporter Assay System (Promega).

For Rluc-Y0×-Fluc and Rluc-Y39×-Fluc reporters, the reaction mixture above was scaled up to 120 μl and incubated at room temperature. 10 μl of the reaction mixture was taken as an aliquot every 5 min and mixed with 20 μl of 1× Passive Lysis Buffer (Promega) to stop the reaction. The Rluc and FLuc luminescence from 30 μl of the mixture was measured with the Dual-Luciferase Reporter Assay System (Promega). The luminescences at 20–30 min were used to calculate the slope of Rluc and Fluc synthesis.

pCI-neo Fluc-EGFP (Addgene plasmid #90170; http://n2t.net/addgene:90170; RRID:Addgene_90170) or pCI-neo FlucDM-EGFP (Addgene plasmid #90172; http://n2t.net/addgene:90172; RRID:Addgene_90172) (kind gifts from Franz-Ulrich Hartl) was transfected into naïve cells or METTL18 KO HEK293T cells by TransIT-293 (Mirus) according to the manufacturer’s instructions. For immunofluorescence staining, cells were cultured on a Nunc Lab-Tek II-CC² Chamber Slide system (Thermo Fisher Scientific) and fixed with 4% paraformaldehyde in PBS for 20 min at room temperature. After being washed twice with PBS containing 0.2% Triton X-100, the cells were incubated with Intercept (TBS) Blocking Buffer (LI-COR Biosciences) with 0.2% Triton X-100 for 1 hr at room temperature. After removal of the blocking buffer, the cells were incubated overnight at 4°C with anti-GFP antibody (Abcam, ab1218) diluted 1:1000 with Intercept (TBS) Blocking Buffer with 0.2% Triton X-100. Then, the cells were washed three times with PBS containing 0.2% Triton X-100 and incubated for 1 hr at room temperature with anti-mouse secondary antibody conjugated with Alexa 488 (Thermo Fisher Scientific, R37120) diluted 1:1000 with Intercept (TBS) Blocking Buffer (LI-COR Biosciences) with 0.2% Triton X-100. Then, the cells were washed three times with PBS containing 0.2% Triton X-100. Slide chambers were mounted with VECTASHIELD HardSet Antifade Mounting Medium (Vector Laboratories). Immunofluorescence images were acquired with a FLUOVIEW FV3000 (Olympus).

To count the aggregation foci, 2 × 10⁵ cells were seeded in 35-mm glass-bottom dishes (IWAKI) and cultured for 24 hr. Cells were transfected with 2 µg of pCI-neo Fluc-EGFP or pCI-neo FlucDM-EGFP and 4 µg of PEI (Polysciences) and cultured for an additional 24 hr. GFP fluorescence was observed under an FLUOVIEW FV3000 (Olympus). At least 100 GFP-positive cells were observed to count cells with aggregates (n = 3).

For MG132 treatment, cells were treated with 0.25 µM MG132 at 24 hr post transfection. After subsequent 24 hr culture, cells were harvested as described in the ‘Ribosome profiling’ section.

300 ng of total RNA purified from cell lysates prepared as described in ‘Ribosome profiling’ section was analyzed with a microchip electrophoresis system (MultiNA, Shimadzu).