METTL1 promotes tumorigenesis through tRNA-derived fragment biogenesis in prostate cancer

Newly growing evidence highlights the essential role that epitranscriptomic marks play in the development of many cancers; however, little is known about the role and implications of altered epitranscriptome deposition in prostate cancer. Here, we show that the transfer RNA N7-methylguanosine (m7G) transferase METTL1 is highly expressed in primary and advanced prostate tumours. Mechanistically, we find that METTL1 depletion causes the loss of m7G tRNA methylation and promotes the biogenesis of a novel class of small non-coding RNAs derived from 5'tRNA fragments. 5'tRNA-derived small RNAs steer translation control to favour the synthesis of key regulators of tumour growth suppression, interferon pathway, and immune effectors. Knockdown of Mettl1 in prostate cancer preclinical models increases intratumoural infiltration of pro-inflammatory immune cells and enhances responses to immunotherapy. Collectively, our findings reveal a therapeutically actionable role of METTL1-directed m7G tRNA methylation in cancer cell translation control and tumour biology. Supplementary Information The online version contains supplementary material available at 10.1186/s12943-023-01809-8.


Introduction
Prostate cancer (PCa) is the second most frequently diagnosed cancer worldwide and the second leading cause of cancer-related deaths in men [1].Conventional therapies target the hormonal axis of the disease; however, up to 30% of patients eventually develop resistance to therapy and metastasis [2], for which limited treatment options exist.With the recent advent of massive parallel sequencing of tumour samples, molecular profiling efforts have revealed a highly diverse genomic, epigenomic, and transcriptomic landscape [3][4][5][6][7], highlighting the need to identify alternative altered and targetable molecular pathways with therapeutic potential.
Growing evidence implicates a key role for non-mutational mechanisms, such as epigenetics and epitranscriptomics, in the survival and growth of cancer cells [8][9][10][11].Over 170 chemical modifications of RNA are known to compose "the epitranscriptome" [12].However, their function remains still widely unknown.In recent years, a growing body of evidence has revealed that aberrant deposition of RNA modifications contributes to cancer progression [11,13,14].
Here, we aimed to identify the RNA modifications associated with PCa progression.Our analysis revealed that overexpression of the RNA methyltransferase METTL1 was linked to PCa.We show that loss of guanosine-7 tRNA methylation is crucial in facilitating the biogenesis of a new class of tRNA-derived small RNAs that favour protein synthesis of a distinctive set of genes, and control cancer cell growth and immune responses.We found that decreased METTL1 expression in human and mouse PCa indicates a better immunotherapy response.
Our study uncovers the complex role of the tRNA m 7 G methylome in PCa and highlights the potential of targeting METTL1 as a novel therapeutic strategy to treat PCa.

METTL1 is elevated in human and murine PCa
To study the potential role of RNA modifications in PCa tumorigenesis, we followed an approach that ensured the selection of relevant RNA modifications associated with PCa tumorigenesis.We focused on identifying alterations in the expression of a set of 132 annotated RNAmodifying proteins (RMPs) (Supplementary Table S1) in datasets from five PCa studies [6,7,[42][43][44] (Fig. 1A).In addition, we extended our analyses to murine PCa using a genetically engineered PCa mouse model [45] (Fig. 1A).We found that the most strikingly differentially expressed gene in primary and metastatic PCa was METTL1 (Fig. 1B; Supplementary Fig. S1A, Supplementary Table S1).METTL1 expression consistently increased from primary to metastatic human tumours (Fig. 1C), and high expression of METTL1 showed worse prognosis in Cambridge, Stockholm, and Taylor cohorts [6,46] (Fig. 1D; Supplementary Fig. S1B).We also found elevated expression of WDR4 (Fig. 1C), the regulatory subunit of the m 7 G RNA methyltransferase complex [30,31], which is also overexpressed in other cancers [22,28,29,38,39]; however, we did not find WDR4 overexpression to be a risk factor for poor prognosis (Fig. 1D; Supplementary Fig. S1C).Increased expression of METTL1 was confirmed in PCa-derived cell lines compared to immortalised benign prostatic hyperplasia or prostatic epithelium-derived cell lines (Supplementary Fig. S1D, S1E).We did not find the same correlation of higher expression of WDR4 in PCa cell lines (Supplementary Fig. S1D, S1E).Increased protein expression of METTL1 and WDR4 was confirmed in PCa samples from a local cohort (Basurto Hospital) (Fig. 1E).Based on the hormonal dependency of PCa tumours, expression analysis of METTL1, WDR4 and AR was performed, but no significative correlation was found between METTL1, WDR4 and AR expression (Fig. 1E).To confirm whether METTL1 and WDR4 expression could correlate with advanced tumour status, we measured PI3K pathway activity by phosphorylation status of S6K, which altered in in approximately 70% of patients with advanced PCa [47].We found a positive and significant correlation between METTL1 and WDR4 expression and PI3K enhanced pathway activation (Fig. 1E), indicating that METTL1 and WDR4 expression is increased in advanced PCa tumours.
In summary, our results demonstrate that METTL1 is the main epitranscriptomic regulator altered in PCa, and its overexpression is correlated with poor prognosis.
Fig. 1 METTL1 is upregulated in prostate cancer.A A schematic overview of the research workflow used to identify altered expression of RMPs associated with PCa.B Heatmap of average Z-scores of mRNA expression values in human primary (P) and metastatic (M) PCa samples compared to healthy tissue for significant differentially expressed RNA-modifying enzymes reveals METTL1 as the most overexpressed RMPs in PCa.Data from are from Grasso et al. [7] (n = N: 12, P: 49, M: 27); Taylor et al. [6] (n = N: 29, P: 131, M: 19); Varambally et al. [42] (n = N: 6, P: 7, M: 6); Lapointe et al. [44] (n = N: 9, P: 13, M: 4); Tomlins et al. [43] (n = N: 23, P: 32, M: 20) (left panel) datasets.Z-score averages for all datasets are also shown as "Aver".Z-score values are indicated using numeric values.Grey Z-score values indicate no significant p-value.The right heatmap shows the average log 2 fold change in mRNA expression values in Pten-cKO mice with prostate intraepithelial neoplasia (PIN) and invasive prostate carcinoma (Inv) compared to normal prostate tissue (right panel) (n = 4).C METTL1 and WDR4 expression are increased in primary (PT) and metastatic tumours (M) compared to normal tissues (N).Data are from Grasso et al. [7], Taylor et al. [6], and Varambally et al. [42] datasets.Log 2 -normalised gene expression values are shown.D High expression of METTL1 but not WDR4 is associated with poor patient prognosis.Kaplan-Meier curves representing biochemical recurrence-free survival (DFS) of patient groups selected according to gene expression, data from the Cambridge, Stockholm and Taylor cohorts [6,46].Data were retrieved from the camcAPP [48] and cBioportal.E Western blot from benign prostatic hyperplasia (BPH) (n = 7) and PCa patient samples (n = 14) (upper panel) and correlation analysis between METTL1 and WDR4 expression, AR and phospho-S6K (right panels).Statistical tests: ANOVA test (B, C), and log-rank Cox test (D).Data are represented as mean ± standard deviation (SD).Student's t-test and Spearman's correlation test (E)

METTL1 expression is regulated by AKT-mTOR downstream signalling pathway
Next, we sought to elucidate the mechanism underlying the upregulation of METTL1 expression in PCa.Because increased androgen receptor (AR) activity is one of the main PCa drivers, we analysed whether increased AR expression or activity, measured by the expression of the receptor and downstream target genes such as KLK3, correlated with METTL1 expression in PCa.In the Grasso dataset [7], we observed an almost significant direct correlation between METTL1 and both AR and KLK3.In the Taylor dataset [6], we found a direct correlation between METTL1 and AR, supporting the notion of a potential relationship between these two factors.However, in the case of KLK3, we did not observe a significant correlation.Our analysis using the TGCA dataset revealed a direct correlation between METTL1 and KLK3 [3].However, interestingly, we did not find a significant correlation between METTL1 and AR in this particular dataset.This inconsistency suggests that the association between METTL1 and KLK3 may vary depending on the dataset being analysed, indicating a lack of robustness or consistency in the observed correlation (Supplementary Fig. S2A).To further clarify whether METTL1 expression was regulated under the AR pathway, we measured the mRNA levels of METTL1 after dihydrotestosterone (DHT) treatment in the androgen-sensitive PCa cell line, LNCaP, and found no changes (Supplementary Fig. S2B), suggesting that METTL1 expression is not regulated by AR activity.Because we found a positive correlation between METTL1 protein expression and phospho-S6K in PCa samples compared to benign prostatic hyperplasia specimens (Fig. 1E), we next analysed whether METTL1 expression was correlated with PTEN expression, a negative regulator of the PI3K/AKT/mTOR pathway, which is deleted in approximately 70% of patients with advanced PCa [47].All datasets analysed showed a significant negative correlation between METTL1 and PTEN expression (Fig. 2A), suggesting that the PI3K-mTOR axis regulates METTL1 expression in PCa.Further dissection of the PI3K-mTORC1/2 pathway using small-molecule inhibitors of PI3K (BKM-120 inhibitor), AKT (MK2206), mTORC1 (rapamycin), and mTORC1/2 (Torin) revealed that AKT inhibition decreased METTL1 mRNA levels, and mTOR inhibitors consistently decreased METTL1 mRNA and protein expression (Fig. 2B-C; Supplementary Fig. S2C-E), indicating that METTL1 expression is regulated via mTOR signalling.
Next, we examined whether METTL1 expression levels helped to define patients with PTEN-loss-associated decreased survival, which is of clinical relevance, and distinguish indolent from aggressive disease in patients with clinically localised tumours [47].Interestingly, we found that high METTL1 expression highlighted a subset of PTEN-low patients with poor outcomes (Fig. 2D).
To ascertain whether METTL1 upregulation in PCa was a direct consequence of low or loss of PTEN expression, we analysed Mettl1 mRNA and protein levels in prostate tissues from wild-type (WT) mice and Probasine-Cre x Pten flox/flox mice (hereafter Pten-KO), with conditionally have deleted Pten in the prostate epithelium.These mice develop high-grade pre-neoplastic lesions after 12 weeks, which progress to invasive adenocarcinoma after 5 months of age [45].We observed a progressive increase in Mettl1 expression after Pten deletion (Fig. 2E-G).Interestingly, tRNAs from mouse prostate Fig. 2 PI3K-AKT-mTORC pathway mediate upregulation of METTL1 expression in PCa.A Correlation analysis between METTL1 and PTEN expression in human primary prostate tumours expression datasets.Plotted values correspond to the log 2 -normalised gene expression values for each patient in the indicated dataset.The black line represents linear regression, grey area indicates the limits of the confidence intervals.Pearson's correlation coefficient (R) and p-values are indicated.Grasso n = 88; Taylor n = 183; TCGA n = 497.B, C METTL1 expression is regulated downstream of AKT signalling.Western blot (B) and RT-qPCR (C) analyses of METTL1 expression upon PI3K pathway inhibition in DU145 cells.DMSO as vehicle (Veh), BKM-120 (BKM) as pan-PI3K inhibitor, MK2206 (MK) as AKT inhibitor, rapamycin (RAPA) as mTORC1 inhibitor, and Torin (TOR) as mTORC1/2 inhibitor.For western blotting, cells were treated for 48 h, and for RT-qPCR, for 8 h.Means ± SD are shown (n = 2) (B) and (n = 6) (C).D Stratification of patients with a worse prognosis according to high METTL1 levels and low PTEN expression.Kaplan-Meier curves representing the disease-free survival (DFS) of patient groups selected according to Q1 (Pten L ) and Q4 (Pten H ) quartile expression of PTEN, and METTL1 high (Met H , log 2 -normalised expression > 8.72) and METTL1 low (Met L , log 2 -normalised expression < 8.72) in recurrent and disease-free (DF) tumour samples from the TCGA dataset.E-G Mettl1 is highly expressed in mouse prostate tumours.Mettl1 expression analysis in Pten-KO mice at 3 and 6 months of age compared to wild-type mice (WT) at the same ages by western blot (E, F) and RT-qPCR (G).* in (E) indicates an unspecific band.Means ± SD are shown for three (F) and five replicates (G).H, J Increased 7-guanine tRNA methylation in RNA from prostate tumours compared with that in normal prostate.North-dot blot of m 7 G levels in long RNAs (> 200 nucleotides long RNAs) and tRNAs extracted from prostatic tissue of Pten-KO mice with intraepithelial prostatic neoplasia (at 3 months of age) or invasive tumour (at 6 months), and wild-type (WT) mice of the same age.Methylene blue staining was used as the loading control (H, bottom panel).Means ± SD are represented (n = 4) (J).I, K Increased 7-guanine tRNA methylation in the urine of mice bearing PCa tumours.North-dot blot of m. 7 G levels from the urines of Pten-KO mice with invasive tumour (at 6 months), and wild-type (WT) mice of the same age.Means ± SD are shown (n > 3).L Representative images of immunostained sections for Mettl1 and markers for luminal (AR), and basal (K14) cells in Pten-KO prostates (anterior lobes) at initiation (3 m)  tumours showed significantly increased deposition of m 7 G in tRNA extracted from tissues, as well as in urine (Fig. 2H-K).Immunohistochemical analysis further revealed that Mettl1 expression was higher in murine PCa and was highly expressed in luminal cells, which are the most common epithelial cell type and are generally accepted as the preferred cell of origin for human prostate cancer [49] (Fig. 2L; Supplementary Fig. S2F, S2G).
Taken together, these data indicate that METTL1 is a downstream effector of the mTORC1 pathway and its activation induces increased expression of METTL1 in PCa.

METTL1 preferentially methylates tRNAs
To understand the role of METTL1 in PCa tumorigenesis, we determined METTL1 RNA substrates by combining two transcriptome-wide methodologies.To identify METTL1-specific RNA targets in PCa cells, we used photoactivatable-ribonucleoside-enhanced cross-linking immunoprecipitation (PAR-CLIP), a stringent technique that incorporates photoreactive ribonucleoside analogues into nascent RNAs that are UV-crosslinked to induce a covalent bond between proteins and RNA, followed by next-generation sequencing [50].We generated PC3 cells expressing doxycycline-inducible HA-METTL1 and used them to immunoprecipitate HA-METTL1 (Supplementary Fig. S3A).Doxycycline-induced cells infected with the empty vector were used as control.We found that tRNAs represented the most enriched RNA species bound to METTL1 when compared to those RNAs bound to the control sample, and with the highest density of reads per gene, with an average median density of Log 2 12 RPMs.(Fig. 3A; Supplementary Fig. S3B; Supplementary Table S2).We did not observed an enrichment of other RNA species bound to HA-METTL1 samples and compared to control samples (Fig. 3A; Supplementary Fig. S3C).To further validate these results, we downloaded the data generated by Bao et al. [51] and analysed the METTL1-bound RNAs in HEK293T cells.tRNAs were also the most abundant RNA species bound to METTL1 in HEK293T cells, with an average median density of Log 2 8 RPMs.The total number of reads mapped to other noncoding RNAs (ncRNAs) and mRNAs was respectively 20% for ncRNAs and approximately 16% for mRNAs, with a median density lower than 16 RPMs per gene, which we considered background (Supplementary Fig. S3D-F; Supplementary Table S3).

METTL1-mediated methylation protects tRNAs from cleavage into small noncoding RNAs
The stability of tRNAs decreases when m 7 G methylation is reduced in yeasts and humans [22,23,27,28,54].To establish that METTL1 loss might perturb the levels of individual METTL1-targeted tRNAs in PCa cells, we performed high-throughput sequencing of the tRNAs isolated from PC3 WT and METTL1 KO cells.In contrast to previous findings, we found no evidence that the loss of METTL1-specific methylation reduces the abundance of specific mature tRNA isoacceptors (Fig. 4A).
Compelling evidence has revealed that the cleavage of a small fraction of tRNAs into tRNA-derived ncR-NAs is a conserved response to stress, and tRNA modifications protect them from stress-induced cleavage [55,58,[61][62][63][64][65].During stress responses, tRNA cleavage was more prominent in METTL1 KO cells than in WT cells, peaking as early as two hours of oxidative stress exposure, and decreasing after 8 h of stress stimulus, probably because of an increase in cell death due to inability to resolve the stress (Fig. 4G, H; Supplementary Fig. S4G, S4J).We did not detect the accumulation of 5′tRNA fragments derived from other METTL1 tRNA targets (i.e.Lys) upon oxidative stress induction (Supplementary Fig. S4H), indicating that 5′tRF formation after stress is tRNA-specific.
In summary, our data indicate that METTL1-mediated methylation is a conserved mechanism that regulates the biogenesis of a novel class of small ncRNAs derived from 5′tRNA fragments during stress responses in PCa cells, regardless their genetic status.

Loss of METTL1 represses translation initiation through tRNA fragment biogenesis
As 5′TOGs are known to repress global translation [57,60,62], we next investigated whether the accumulation of 5′TOGs contributed to translation alterations in PC3 METTL1-depleted cells.We confirmed a reduction in protein translation measured by O-propargyl-puromycin (OP-puro) incorporation in METTL1 KO cells compared to that in WT cells (Fig. 5A).The changes in protein synthesis rates were not associated with expression changes of components of the translation initiation complex or in the phosphorylation status of eIF2α (Supplementary Fig. S5A).
The production of 5'tRNA fragments in response to cellular stress play a critical role in the stress response mechanism.This process helps to inhibit global protein synthesis, allowing the cells to recover and survive stressful conditions [66].To determine whether the removal of METTL1 impacts the stress response, we measured protein synthesis rates after oxidative stress induction.Our analysis showed that METTL1-deficient cells displayed a reduced recovery rate of protein synthesis when compared to WT cells, indicating that METTL1 KO cells are more sensitive to stress (Supplementary Fig. S5B).This finding suggests that the loss of m 7 G significantly repressed protein synthesis and this process is independent of eIF2α phosphorylation status.
Next, we sought to define the molecular basis for the distinctive translational repression of METTL1 KO cells and the dependence on 5′TOGs.Previous evidence has shown that 5′TOGs impair translation initiation by displacing the components of the translation initiation complex eIF4A/G/E and regulatory factors YB1 and PABP1 from the cap of mRNAs [57,60,67].To determine whether translation initiation factors were displaced from 7-methyl-guanylated (m 7 G)-capped mRNAs in PC3 METTL1 KO cells, we analysed the affinity of translation initiation factors for m 7 G-cap-coated sepharose beads.We found that the m 7 G-caps affinity of eIF4G and PABP1 was significantly decreased in METTL1 KO cells compared to that in WT cells (Fig. 5B; Supplementary Fig. S5C).This suggested that the increased accumulation of 5′TOGs destabilised the assembly and binding to the cap of some factors of the translation initiation complex, leading to repressed translation in METTL1 KO cells.
We further evaluated whether synthetic 5′TOGs expressed in WT cells were able to bind to translation initiation and regulator factors and whether their affinity to these factors could be blocked using synthetic reverse-complement 5′TOG RNAs (or anti-TOGs).To this end, we transfected PC3 WT cells with 5′biotinylated-5′TOG RNAs (containing the sequence of the most abundant 5′TOG found in PC3 METTL1 KO cells) with or without the addition of anti-TOGs.After pulldown of 5′biotinylated-5′TOGs, we found that the affinity of PABP1 and YB1 for 5′TOG was significantly reduced in the presence of anti-TOGs (Fig. 5C; Supplementary Fig. S5D).Next, we investigated whether synthetic 5′TOGs could phenocopy the translation initiation complex displacement in WT cells and whether anti-TOG RNAs could rescue this observed effect in METTL1 KO cells.We found that the binding of PAPB1 to m 7 G-caps was displaced in WT cells transfected with 5′TOGS, but the m 7 G-caps affinity of PAPB1 was significantly increased in METTL1 KO cells transfected with anti-TOG RNAs (Fig. 5D; Supplementary Fig. S5E).Altogether indicating that the formation of 5′TOGs repress protein translation by displacing PAPB1.Interestingly, PAPB1 has been previously shown to have a strong affinity for 5′TOGs in other cell types [57].
Thus, our results confirm that 5′TOGs expressed in METTL1 KO cells can displace translation regulatory factors from mRNA caps and provide key insights into the molecular basis underlying METTL1-mediated translational repression.

METTL1 downregulation suppresses prostate tumour growth in vitro and in vivo
Given that METTL1 KO cells synthesise fewer proteins, we hypothesised that METTL1 inhibition could reduce cell and tumour growth.In fact, we observed that knocked down of METTL1 impaired the growth of PC3, DU145 and 22Rv1 cells (Fig. 5E; Supplementary Fig. S6A, S6B).METTL1 downregulation also diminished cell division, induced cell cycle arrest, increased apoptosis, impaired spheroid formation capacity, and drastically reduced the growth and proliferation of tumour xenografts (Fig. 5F-I; Supplementary Fig. S6C-F).Thus, our data indicated that METTL1 downregulation can effectively decrease tumour growth, strongly supporting that METTL1 is crucial in the regulation of PCa progression.
Next, we investigated whether METTL1 methylase activity was sufficient to promote cell growth.We reexpressed a wild-type version of METTL1 (WT) and a catalytically dead mutant (AFPA) in PC3 METTL1 KO cells (Fig. 5J-L; Supplementary Fig. S6G-I).The proliferation and spheroid formation capacity of METTL1 KO cells re-expressing WT METTL1 were increased compared to KO cells transduced with an empty vector (eV).In contrast, the expression of the catalytically inactive mutant failed to promote METTL1 KO cell growth and spheroid formation capacity (Fig. 5M-N; Supplementary Fig. S6J, S6K).Because METTL1 catalytic activity is supported by the formation of a complex with the regulatory subunit WDR4 [30,31], and WDR4 is overexpressed in PCa (Fig. 1C, E), we tested the oncogenic potential  S5 (B, C, D).E Growth curves of PC3 METTL1 KO, WT, and parental cells (PC3).Mean ± SD, n = 3.The dotted line represents the average growth of WT and parental or KO cells.F Reduced cell division rate, as measured by BrdU incorporation.G METTL1 depletion increased apoptosis in PC3 METTL1 KO cells.Flow cytometry analysis of Annexin V staining.Mean ± SD, n = 3. H Reduced spheroid formation capacity in PC3 METTL1 KO cells.Means ± SD, n = 3.The dotted line represents the average values of all WT or KO cells.I Tumour growth in xenografted PC3 METTL1 KO and WT cells in athymic nude mice reflects impaired tumour formation in the absence of METTL1.Mean ± SEM, n = 10.J-L Protein expression (J) and m 7 G methylation levels of tRNAs (K, L) of PC3 METTL1 KO cells ectopically expressing a doxycycline-inducible HA-tagged wild-type (WT) or a catalytic dead mutant (AFPA) version of METTL1.PC3 METTL1 KO cells were infected with an empty vector (eV) as a control.Methylene blue staining was used as the loading control (L, bottom panel).Mean ± SD, n = 3 (L).M, N Proliferation (M) and spheroid formation capacity (N) were dependent on METTL1 catalytic activity.PC3 METTL1 KO cells re-expressing METTL1 (WT) or catalytic dead mutant (AFPA) compared to METTL1 KO cells infected with empty vector (eV).Mean ± SD, n = 6.O, P 5'TOG transfection induces apoptosis and reduces cell proliferation.Percentage of apoptotic (O) and growth rates (P) of PC3 METTL1 KO and WT cells after transfection with synthetic 5'TOGs (TOG) or anti-TOG (ANT) RNAs.Controls (Cont) were transfected with scramble RNAs.Mean ± SD, n = 6 (O), n ≥ 10 (P).Statistical tests: Two-way ANOVA (E, I, L), one-way ANOVA (F, G, H), and one-tailed Student's t-test (A-D, M-P).*p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001 of WDR4.Inducible silencing of WDR4 did not reduce cell or tumour growth (Supplementary Fig. S6L, S6M).Together, these results suggest that METTL1 promotes the growth of PCa cells in an enzymatic activity-dependent manner, but is independent of WDR4.
To further determine whether 5′TOGs were sufficient to induce growth arrest and apoptosis, we transfected PC3 WT cells with 5′TOG and METTL1 KO cells with anti-TOG and measured apoptosis and proliferation.We detected a significant increase in apoptosis and decrease in proliferation when WT cells were transfected with 5′TOG RNAs, and a decrease in apoptosis, although not significant, and a significant increase in proliferation when METTL1 KO cells were transfected with anti-TOGs (Fig. 5O, P).
Altogether, our data suggest that METTL1 inactivation and 5′TOGs are required to induce growth arrest, and the re-expression of the catalytically active version of METTL1 and anti-TOGs can partially rescue the observed defects in METTL1 KO cells.

METTL1 loss activates IFN signalling pathway
Given that translation is affected by METTL1 inhibition, we explored the consequences of METTL1 loss on immediate translatome changes.To this end, we took advantage of the OP-puro properties to label nascent proteins that were subsequently conjugated to biotin-coated beads, followed by on-bead digestion and LC-MS/MS (Fig. 6A) [68].Gene ontology (GO) term enrichment analyses identified decreased translation of genes linked to cell division and mitosis, corroborating the observed defects in proliferation of METTL1 KO cells (Fig. 6B; Supplementary Table S5).Unexpectedly, we also found increased translation of transcripts associated with stress responses, including type I and II interferon (IFN) signalling pathways, immune effector processes, and catabolic processes in METTL1 KO cells (Fig. 6B).The differences in translation were reflected in the global proteome composition of PC3 METTL1 KO cells, but did not correlate with RNA expression levels, suggesting that post-transcriptional or translational regulation was responsible for the uncorrelated RNA-protein expression levels found in METTL1 KO cells (Fig. 6C; Supplementary Fig. S7A, S7C, S7D; Supplementary Tables S6, S7, S8).To further test whether the translational efficiency of certain transcripts differed in METTL1 KO cells, we performed polysome profiling.This approach enabled us to confirm the decreased formation of active polysomes in METTL1 KO cells, concomitant with a reduction in global protein synthesis (Fig. 6D).Analysis of mRNA enrichment in the polysome fractions of METTL1 KO cells revealed increased translation of specific transcripts linked to interferon signalling, such as interferon regulatory factor 9 (IRF9) and interferon-stimulated gene 15 (ISG15) (Fig. 6D).As a control for specificity in translational changes, we did not find enrichment of transcripts such as GAPDH, Catenin β, KIF20A or STAT1 in the polysome fractions of METTL1 KO cells (Fig. 6D; Supplementary Fig. S7B).Taken together, our findings indicate that METTL1-mediated tRNA methylation steers a distinct translational programme.
Because 5′TOGs can re-programme the translation machinery to favour the translational programmes required for cancer cells [58,69], we next investigated whether translational changes in METTL1 KO cells were mediated by the increased biogenesis of 5′TOGs.To this end, we transduced PC3 WT and METTL1 KO cells with 5′TOG and anti-TOG RNAs and assessed the changes in protein expression.We found that 5′TOGs transfection in WT cells increased IRF9 and ISG15 protein expression, whereas transfection of anti-TOG RNAs in METTL1 KO cells slightly but significantly decreased the expression of both proteins (Fig. 6E; Supplementary Fig. 6 Loss of m 7 G tRNA methylation results in distinct translational programmes.A Schematic overview of nascent polypeptide OP-puro-labelling and enrichment followed by LC-MS/MS peptide identification and quantification analysis (upper panel).Lower panel shows common differentially expressed nascent proteins ranked in a volcano plot according to their statistical p-value (-Log 10 pV) and their relative abundance ratio (Log 2 FC) between four replicates of PC3 WT and METTL1 KO cells.Coloured dots represent statistically (p-value < 0.05) upregulated (red) and downregulated (blue) proteins in METTL1 KO cells.B Gene Ontology (GO) category enrichment of biological processes in significantly (p-value < 0.05) upregulated (UP, FC > 1) or downregulated (Down, FC < 1) nascent proteins in METTL1 KO cells compared to WT cells.Categories were ranked according to their statistical P-value (-Log 10 pV) and fold enrichment of genes found for each category.C No correlation between differentially expressed proteins (protein log 2 FC) and their mRNA (protein log 2 FC) was observed in PC3 WT vs. METTL1 KO cells.Coloured dots represent significant (p-value < 0.05) differentially expressed proteins (blue), mRNAs (yellow) or both (red) for each gene.Fig. S7E).We also observed increased protein expression and phosphorylation of STAT1 in METTL1 KO cells and in xenografts (Fig. 6E; Supplementary Fig. S7E, S7F), yet its increased expression was independent of increased translation (Fig. 6D), but was transcriptionally dependent (Supplementary Fig. S7G), suggesting that STAT1 increased expression was induced by activation of the IFN signalling pathway in METTL1 KO cells [70,71].In fact, the transcription of other interferon-stimulated genes (ISGs) was also transcriptionally activated in METTL1 KO cells (Supplementary Fig. S7G).We also observed that transient silencing of METTL1 in another cell line, 22Rv1, significantly increased the expression of ISGs genes, most likely due to a weak but not significant increased expression of IRF9 (Supplementary Fig. S7H, S7I).Given the anti-proliferative and pro-apoptotic effects of STAT1 activation [72,73], we tested whether the growth defect in METTL1 KO PC3 cells could be reversed by downregulating STAT1.We found that apoptosis was reduced and proliferation was increased in PC3 METTL1 KO cells after STAT1 knockdown (Fig. 6F, G), suggesting that the antiproliferative effect of METTL1 inhibition can be in part mediated by activation of the STAT1 signalling pathway.
Taken together, our data suggest that METTL1-mediated tRNA methylation and tRNA fragment biogenesis induce a translational programme that activates the IFN signalling pathway.To confirm that high METTL1 expression correlates with a decrease in IFN pathway activity in human PCa samples, we assessed ISGs expression in three PCa expression datasets [3,6,7].We observed a significant inverse correlation between METTL1 and STAT1 expression (Supplementary Fig. S7J).Similarly, we found an inverse correlation, although weaker, between METTL1, IFIT1 and IFIT2 expression (Supplementary Fig. S7J).
In summary, our data indicate that METTL1 expression inversely correlates with activation of the IFN pathway in PCa cells and tumours, and METTL1 inhibition translationally activates the IFN signalling pathway.

Low METTL1 expression in PCa correlates with increased pro-inflammatory immune cell polarisation
Recently, epigenetic targeted therapies have been shown to trigger an IFN antiviral response in several cancer types, including PCa, mounting an innate immune response and resulting in the production of many cytokines [74][75][76][77].To elucidate whether METTL1 inhibition can mount an innate immune response by activating the IFN signalling pathway, we investigated whether METTL1 downregulation could alter cytokine expression and secretion in PCa cells.Overall, in METTL1 KO cells, the cytokine composition analysis revealed an increased secretion of cytokines involved in pro-inflammatory activity and which polarise macrophages to M1-like endotype [78], including granulocyte-macrophage colony-stimulating factor (GM-CSF) and tumour necrosis factor α (TNF-α) (Fig. 7A; Supplementary Fig. S8A).METTL1 removal also induced downregulation of antiinflammatory cytokines, including Macrophage or Colony stimulating factor (M-CSF), IL10 and IL13 (Fig. 7A; Supplementary Fig. S8A), which can polarise macrophages to M2-like endotype [78].These data suggested that METTL1 inhibition in tumour cells could polarise immune cells in the tumour microenvironment (TME) towards a cytotoxic tumoricidal endotype.
The advent of molecular technologies, such as singlecell RNA sequencing (scRNA-seq) and spatial transcriptomics, has enabled unprecedented auditing of the composition of the tumour immune microenvironment.Unsupervised clustering of scRNAseq and spatial transcriptomics analysis has revealed that within the innate immune population, macrophages constitute the largest proportion in PCa [79][80][81].Tumour-associated macrophages (TAMs), which typically exhibit an expression pattern characteristic of M2-like or pro-tumoural macrophages [82], represent a major component of the TME supporting tumorigenesis [83].They are known to facilitate cancer cell proliferation, invasiveness, angiogenesis, and immunotolerance, while also contributing to resistance against standard-of-care therapies, immunotherapies, and chemotherapy [84][85][86].Consequently, there is a critical need to develop clinical agents that can either deplete macrophages or induce their repolarisation toward an M1-like or anti-tumorigenic state [84,87].
Taking these circumstances into account, along with the observation that METTL1-expressing cells secrete cytokines known to drive macrophage polarisation towards the M2-like endotype, we focused on elucidating the role of METTL1 in macrophage polarisation.We incubated the human monocyte cell line THP1 with conditioned media (c.m.) from PC3 WT or METTL1 KO cells and investigated the expression of M1-like and M2-like macrophage endotype markers (Fig. 7B).tSNE analysis showed that while c.m. from WT cells induced macrophages towards an M2-like endotype, inhibition of METTL1 in PCa cells induced macrophages towards an M1-like endotype (Fig. 7B; Supplementary Fig. S8B).Macrophages in the presence of METTL1 KO cells' c.m. also showed a phagocytic capacity similar to that of M1-like macrophages (Supplementary Fig. S8C).Similar results were obtained when primary peripheral blood monocytes were exposed to PC3 METTL1 KO cells' c.m. (Supplementary Fig. S8D).In addition, peripheral blood macrophages cultured in the presence of PC3 METTL1 KO cells' c.m. showed higher induction of the proliferation and migration of CD3+ T cells (Fig. 7C, D; Supplementary Fig. S8E).Altogether, our data suggested that METTL1 inhibition in cancer cells may stimulate a cytotoxic and anti-tumoural inflammatory response in the TME.
To validate our findings in vivo, we evaluated the intratumoural immune cell composition of human prostate tumours using immunohistochemical analyses.The analysis showed a significant direct correlation between M2-like macrophage infiltration and METTL1 expression at the expenses of an inverse trend in M1-like macrophage infiltration in PCa paraffin samples.Similarly, CD8+ T cells infiltration correlated inversely with METTL1 expression (Fig. 7E; Supplementary Fig. S8F).Additionally, deconvolution analysis of PCa expression datasets using CIBERSORT [88] or TIMER [89] showed a significant inverse correlation between intratumoural M1-like macrophage composition and METTL1, indicating a more cytotoxic immune tumour component in METTL1-low tumours (Supplementary Fig. S8G, S8H).
Altogether, our results demonstrated that inhibition of METTL1 expression in PCa cells can mount a cytotoxic immune response in vitro and further support that low expression of METTL1 in human PCa is associated with an increase in intratumoural cytotoxic immune cells in the TME.

Mettl1 inhibition enhances the response to immune checkpoint blockade therapy in PCa
To assess whether METTL1 inhibition could reduce the tumour growth and affect the intratumoural immune composition in vivo, we deleted Mettl1 in Pten-KO mice (Supplementary Fig. S9A, S9B).Mettl1 systemic homozygous deletion was lethal (data not shown).Mettl1 heterozygous deletion (Pten-KO/Mettl1 +/-) and Mettl1 conditional deletion in the prostatic epithelium of Pten-KO mice (Pten-KO/Mettl1 flox/flox ) resulted in a significant decrease in the tumour volume of the ventral and anterior lobes, which were consistently larger in Pten-KO/Mettl1 + / + mice (Fig. 7F; Supplementary Fig. S9F).Systemic and conditional knockdown of Mettl1 in Pten-KO mice significantly increased the formation of 5′tRNA fragments, indicating that tRNA cleavage protection by m 7 G methylation is a conserved mechanism in mice (Supplementary Fig. S9C).Histological analyses of tumours showed that, while Mettl1-expressing tumours were invasive and consisted of undifferentiated cells, tumours from Pten-KO/Mettl1 +/and Pten-KO/ Mettl1 flox/flox mice were encapsulated and consisted of differentiated luminal cells with low proliferative potential (Supplementary Fig. S9D, S9E, S9G), suggesting that Mettl1 is required for tumorigenesis.
To investigate whether the mechanism underlying the anti-tumour effect of Mettl1 deletion was due to tumour cell intrinsic pathways or to the tumour immune microenvironment, we injected tumour cells isolated from Pten-KO/ Mettl1 + / + and Pten-KO/Mettl1 fl/fl mouse prostates into NOD scid gamma (NSG) immune-incompetent mice.Allografts growth revealed a slight but non-significant reduction in the growth of allografts derived from Pten-KO/Mettl1 fl/ fl mice (Supplementary Fig. S9H).These findings provided important insights into the mechanism underlying the antitumour effect of Mettl1 deletion and suggested that other than tumour cell intrinsic pathways were involved in the anti-tumour effects observed upon Mettl1 deletion in vivo.
Based on these observations and the capability of METTL1-deficient cells to elicit a cytotoxic immune response characterised by the induction of M1-like macrophage polarisation, as well as enhanced proliferation and migration of CD8+ T cells in vitro (Fig. 7A-D, Supplementary Fig. S8A-E), our subsequent investigation focused on analysing the composition of macrophages and CD8+ T cells in Mettl1-deficient prostate tumours.Immunohistochemical analysis revealed a significant increase in intratumoural infiltration of M1-like macrophages (iNOS+) at the expense of M2-like macrophages (Arg1/CD68+), and an increase in CD8 + T cells in Mettl1 +/and Met-tl1 flox/flox tumours compared to Mettl1 +/+ tumours (Fig. 7G; Supplementary Fig. S9E, S9G).
Cytotoxic T cells, distinguished by their expression of CD8, possess the remarkable ability to directly recognise and eliminate cancer cells, making them the most potent effectors in the anticancer immune response [90].To determine whether the observed anti-tumour effect upon Mettl1 deletion in mice was due to increased intratumoural infiltration of cytotoxic CD8+ T cells, we examined the effect of systemic elimination of CD8+ T cells in Pten-KO/Mettl1 + / + and Pten-KO/Mettl1 fl/fl mice treated with anti-CD8α antibodies.Our results showed that the elimination of CD8+ T cells led to a significant increase in the size of only anterior lobe tumours lacking Mettl1 (Supplementary Fig. S9I).These results suggested the partial involvement of CD8+ T cell infiltration in the anti-tumour effect observed in Mettl1-deficient tumours, however, it also indicated the involvement of alternative immune cell types or pathways.
Cancer-related inflammation plays a crucial role in the progression of most cancer types.Components of both the innate and the adaptive immune system that are recruited upon tumour onset interact with each other and with the tumour, thus building an immune response in situ.In prostate tumours, various types of cytotoxic immune cells have been observed to infiltrate the TME [79,81].Dendritic cells (DCs), are crucial antigen presenting cells that present MHC-peptides to cytotoxic T cells, play a vital role in tumour immunotherapy [91].Natural killer (NK) cells are an innate lymphoid cell with cytotoxic function and can directly kill tumour cells without prior sensitisation [92].The adaptive immune system, composed of CD8+ T and also CD4+ T cells, and B cells [79], plays a pivotal role in mounting an effective, antigen-specific immune response against tumours.Cytotoxic T cells for instance are key effectors of the adaptive immune system and can directly recognize and eliminate cancer cells [90].Substantial evidence has shown that cancer cells can orchestrate this pro-tumourigenic response by reprogramming the immune system.Thus, future functional validations showing how METTL1expressing tumour cells interact with their neighbouring immune cells will lead to a better understanding of PCa progression.
To elucidate the impact of Mettl1 inhibition on the reprogramming of the immune tumour composition, we employed an analysis of immunomodulatory molecules, including cytokines and chemokines, as surrogate markers of the immune cells present in the tumour.Our results revealed a significant shift in the cytokine composition, characterised by a decrease in the secretion of pro-tumoural cytokines.Additionally, we observed an increased secretion of cytokines known to correlate with a favourable response to immune checkpoint blockade (ICB) therapy, in Pten-KO/Mettl1 fl/fl tumours (Fig. 7H) [93].These findings suggest that Mettl1 inhibition may contribute to an immune microenvironment conducive to improved ICB therapy outcomes, highlighting the potential of targeting Mettl1 as a therapeutic strategy for enhancing the anti-tumour immune response.
To determine if the cytokine intratumoural composition in Mettl1-deficient tumours could enhance the efficacy of ICB therapy, we treated mice with anti-PD1 and anti-CTL4A antibodies.Pten-KO/Mettl1 + / + mice did not shown tumour volume reduction after ICB treatment compared with that in untreated mice, whereas Pten-KO/ Mettl1 flox/flox mice exhibited a significant reduction in tumour volume after ICB treatment (F ig.7I; Supplementary Fig. S9J).Remarkably, the tumour volume of Pten-KO/Mettl1 flox/flox mice after ICB treatment was reduced to almost the size of the normal prostate (Supplementary Fig. S9J).Therefore, the results strongly suggested that decreased expression of Mettl1 increased the efficacy of ICB therapy in preclinical mouse models.
We also investigated whether METTL1 expression levels could predict the efficacy of ICB therapy in patients.Using the ROC plotter platform, we analysed the expression levels of METTL1 in several tumour types treated with anti-PD1 therapy [94].We found that METTL1 expression was lower in responders than in non-responders to ICB therapy in breast, colorectal, ovarian cancer, and glioblastoma patients, suggesting that high METTL1 expression predicted poor response to ICB therapy (Fig. 7J).
In summary, our findings demonstrate that METTL1 expression levels can determine intratumoural cytotoxic immune infiltration and the success of ICB therapy.

Discussion
RNA modifications are emerging as important modulators of biological processes and pathological progression.Although recent studies have shown that their dysregulation is associated with cancer, we are in the infancy of our understanding of the molecular basis.In particular, tRNA modifications are emerging as important regulators of cancer; however, molecular insights remain poorly understood [95].In this regard, our study identifies that increased expression of the tRNA methyltransferase METTL1 and elevated 7-methylguanosine deposition in tRNAs have oncogenic potential by altering the translational programme in cancer cells.In line with our findings, recent evidence has associated increased METTL1 expression with the occurrence of several cancer types [22, 23, 28, 29, 32-34, 36, 37, 96, 97].In addition, we found that increased expression of Mettl1 correlates with increased m 7 G deposition in tRNAs and increased levels of excreted 7-methylguanine in the urine in mouse PCa.Increased m 7 G excretion has been found in other tumour-bearing rodents [98] and in patients with cancer, including PCa [99][100][101][102].This evidence suggests that both METTL1 expression and m 7 G deposition levels may serve as potential tumour biomarkers [103].
PTEN-PI3K pathway is among the most commonly altered pathways in advanced PCa and is a critical indicator of disease progression in hormone-naive and castration-resistant prostate cancer (CRPC) [47].Thus, mechanistic understanding and identification of the downstream effectors of the PI3K-PTEN oncogenic signalling pathway can lead to better patient stratification and the design of better therapeutic strategies.Here, we provide unprecedented evidence of the control of tRNA methylation downstream of the AKT pathway through the control of METTL1 expression.Consistent with our findings, METTL1 was reported to be regulated by AKT and ribosomal S6 kinase in embryonic stem cells [104].In hepatocellular carcinoma, METTL1 expression has been associated with PTEN signalling [33].Moreover, in lung adenocarcinoma cell lines, a positive feedback loop has been suggested, in which METTL1 expression enhances the AKT-mTOR signalling pathway [97].These findings highlight the complex regulatory landscape that controls tRNA modifications and integrates and translates growth signals into oncogenic epitranscriptomic programmes.
Our study provides strong evidence supporting an oncogenic role for METTL1-mediated tRNA methylation in PCa.We found that the deletion of METTL1 dramatically suppressed tumorigenesis both in vitro and in vivo.In contrast, re-expression of only a catalytically active version of METTL1 promotes the proliferation of cancer cells, suggesting that METTL1-mediated RNA methylation has oncogenic potential.Consistent with our functional in vitro and in vivo data on PCa, recent studies have highlighted the oncogenic role of METTL1 in several cancer types [22,23,28,33,37,97].Mechanistically, we found that METTL1 primarily methylates tRNAs to protect them from endonucleolytic cleavage, thereby determining the production and identity of a subset of tRNA fragments (tRFs) known as 5′TOGs.The production of tRFs is a common response to various stimuli, including stress, and their biogenesis is regulated by RNA modifications [58].In this regard, tRNA fragment diversity and abundance are particularly associated with cancer [105], with PCa showing the highest repertoire [106,107].The molecular role of tRNA cleavage is not to degrade hypomodified tRNAs; instead, 5′TOGs are specifically produced to reprogramme mRNA translation [16,55,57,60,67].Here, we found that in METTL1-deficient cells, 5′TOGs displace translation initiation factors from capped-mRNAs to favour the translation of a distinct set of genes involved in tumorigenesis and stress responses.Recent studies have shown that METTL1 regulates self-renewal, differentiation, and growth through different mechanisms related to codon-specific translational control [22,23,27,28].Thus, our study uncovered new layers of selective regulation of gene expression that cancer cells selectively utilize to favour survival and growth, independent of transcriptional control.
Protein synthesis rates and translational programmes are highly regulated in homeostasis, and dysregulation of these translation programmes is associated with cancer [108,109].In PCa, the PI3K-mTOR axis steers cancer pathogenesis by regulating alternative translation initiation factors that favour distinct translational programmes [110].In this regard, our findings support a model whereby tRFs biogenesis is controlled by epitranscriptomic regulation to reprogramme the translatome to allow cell growth and adaptation to the TME [16,55,57,59].
In our study, we have identified METTL1, by modulating mRNA translation, is a crucial regulator of the IFN-STAT1 signalling pathway in tumour cells, implying its significance in prostate cancer and the tumour microenvironment.Our findings indicate that inhibiting METTL1 activity leads to favourable changes within the tumour, such as an increased presence of anti-tumoural cytokines and the infiltration of cytotoxic immune cells, including M1-like macrophages and CD8+ T cells.The results of our research provide compelling evidence highlighting the pivotal role of METTL1 in regulating the immune response against tumours.This emphasises its potential as a promising target for novel cancer therapies, either as a standalone treatment or in combination with immune checkpoint blockade therapy.In line with our observations, recent studies have shown that METTL1 is involved in mediating the mRNA translation of cytokines that induce the accumulation of myeloidderived suppressor cells (MDSCs) in intrahepatic cholangiocarcinoma [111], a major immune suppressive subset infiltrating these tumours [112].Moreover, our study aligns with previous evidence demonstrating the association between METTL1 dysfunction and chronic immune conditions in humans, such as psoriasis and multiple sclerosis [113][114][115][116].It will be interesting in future studies to explore the immune landscape in the context of METTL1 to gain a more comprehensive understanding of the implications for tumour immune evasion in PCa other cancer types.In conclusion, our study underscores the critical role of METTL1 in PCa and the tumour microenvironment.The insights gained from this research open up new possibilities for targeted therapies that modulate METTL1 activity, which may enhance the anti-tumour immune response and improve patient outcomes in PCa.These findings also highlight that the impact of METTL1 extends beyond cancer and may have implications in broader immune-related disorders.
Central to the efficacy of ICB therapy is the requirement for cytotoxic immune cells to infiltrate the tumours [117].ICB therapy generates durable therapeutic responses in a significant subset of patients across a variety of cancer types [118].However, PCa has shown resistance to ICB owing the high infiltration of immunosuppressive myeloid cells [79][80][81][119][120][121][122].In addition, suppression of the tumour-intrinsic IFN pathway is strongly associated with tumour cell outgrowth and metastasis in PCa [123].Emerging evidence has demonstrated that enhancing immune responses and reprogramming or re-education of intratumoural immune cells towards a tumoricidal endotype is an effective strategy to boost ICB therapy to treat cancer [124][125][126].Epigenetic inhibitors are emerging as immune modulators of the TME.Epigenetic inhibitors can induce IFN pathway activation [74,127], re-educate TAMs towards an M1-like endotype, and increase intratumoural cytotoxic T cell infiltration, all resulting in superior immune therapeutic benefits when combined with ICB [75,[128][129][130][131][132].The therapeutic potential of targeting the epitranscriptome in cancer remains largely unexplored, but recent research suggests that the epitranscriptome may play a role in immune regulation [133][134][135].Our study contributes to this understanding by demonstrating that inhibiting METTL1 leads to enhanced infiltration of cytotoxic macrophages and cytotoxic T cells into tumours, indicating that METTL1 serves as a key mediator of intratumoural immune responses.These findings suggest that combining METTL1 inhibitors with immunotherapy could have a synergistic effect, increasing the efficacy of cancer treatment.

Conclusions
Our study provides compelling evidence of MET-TL1's oncogenic function and overexpression in prostate cancer (PCa).We also discovered a new layer of expression regulation and selective translation control through tRNA fragment biogenesis.By inhibiting METTL1-mediated methylation, we were able to increase the biogenesis of a novel class of ncRNAs (5'TOGs), which regulate the activation of the interferon signalling pathway in cancer cells.Additionally, METTL1 inhibition in cancer cells leads to increased infiltration of cytotoxic immune cells, which transforms the immunosuppressive prostate TME into a tumoricidal endotype.These results suggest that targeting METTL1 alone or in combination with immune checkpoint inhibitors holds great potential for developing effective therapeutic strategies, particularly for patients with castration-resistant prostate cancer who have limited treatment options.

Mouse lines
All mice were housed at the Animal Research Core Facility at CIC-bioGUNE and at the University of Salamanca and maintained in ventilated filter cages under specific pathogen-free conditions with food and water available ad libitum.All mouse experiments were carried out following the ethical guidelines established by the Biosafety and Bioethics Committee at CIC bioGUNE (under protocol P-CBG-CBBA-0715), and at the University of Salamanca (under protocols #269, #506, #595, #707), and by the Competent Authority of the Basque Country and Castilla y León.Xenograft experiments were performed by subcutaneously injecting 1 × 10 6 PC3 WT and METTL1 KO cells with Matrigel (BD Biosciences) in two flanks per mouse of eight weeks old athymic nude mice (Envigo).pLKO-Tet-on-shRNA-WDR4-2 or pLKO-Tet-on-shRNA-SCR cells were injected into each flank of athymic nude mice and when tumours were growing, mice were fed normal or doxycycline diet (Research diets, D12100402) to induce silencing.Mice in this group were left on continuous doxycycline diet.Cells were pretreated for 48 h with PBS or doxycycline (0.1-0.5 μg/ ml).Allograft experiments were performed by injecting 1 × 10 6 tumour cells isolated from five months old Pten-KO/Mettl1 + / + and Pten-KO/Mettl1 flox/flox mice.To isolate cells from Pten-KO/Mettl1 + / + and Pten-KO/ Mettl1 flox/flox mice, tumours were dissected and dissociated for 1 h at 37ºC in rotation with DMEM-F12 containing 5% FBS and 1 × collagenase Type 1 (Ther-moFisher) and 1 × hyaluronidase (Stem Cell Tech.).Next, cells were incubated for 1 h at 37ºC with 0.25% trypsin, followed by incubation with 5 mg/ml Dispase II (Sigma-Aldrich) and 1 U/µl DNase (Sigma-Aldrich) for 30 min at 37ºC.Dissociated cells were then filtered through a 0.45 µm cell strainer and mixed 1:1 with Matrigel (BD Biosciences) and injected subcutaneously into both flanks of NSG (NOD.Cg-Prkdc scid Il2rg tm1Wjil /SzJ) mice (Charles River).Tumour were grown to a maximum total diameter of 1.2 cm of diameter.Tumour volume was estimated using the following formula: volume = (length x width) 2 × 0.526.
To generate a conditional and straight Mettl1 mouse model, we floxed exon 2 of Mettl1 at a residue overlapping the splicing site (Supplementary Fig. S9A).In silico analysis revealed that an early truncated METTL1 protein without catalytic activity was produced when exon 2 was removed.Two complementary RNA oligos corresponding to crRNAs (CRISPR RNA), named METTL1 crRNA guide 1 and guide 2 (Mettl1 crRNA guide1: rU rA rA rG rA rG rC rC rA rU rG rA rU rG rA rUr C rC rA rA rG rU rU rU rU rA rG rA rG rC rU rA rU rG rC rU; Mettl1 crRNA guide2: rU rG rG rC rU rU rG rU rU rA rG rG rU rA rA rU rA rA rG rC rG rU rU rU rU rA rG rA rG rC rU rA rU rG rC rU), were designed with the Spanish National Biotechnology Centre (CNB)-CSIC web tool (http:// bioin fogp.cnb.csic.es/ tools/ break ingcas/).These crRNAs contained the target-specific sequence for guiding Cas9 protein to Mettl1 exon 2 and the adjacent The crRNAs, ssODN, and trans-activating CRISPR RNA (tracrRNA) were produced by chemical synthesis at Integrated DNA Technologies (IDT).The crRNA and tracrRNA were annealed to obtain the mature sgRNAs.A mixture containing sgRNAs, recombinant Cas9 protein (IDT), ssODN [Cas9 (30 ng/µL), annealed sgRNA (20 ng/µL each), and ssODN (10 ng/µL)] was microinjected into C57BL6/J zygotes at the Transgenic Facility (NUCLEUS, platform for research support of the University of Salamanca).All edited founders were identified by polymerase chain reaction (PCR) amplification (Taq polymerase, NZYTech) with primers flanking exon 2 (primer F 5'-TCT CCT GGT GTG CAT GAA GAC-3' and R 5′-AGG GAA GGT GGT GGA ATC CC-3′), producing an amplicon of 957 base pairs (bp) for edited alleles with a new enzyme restriction site.PCR products from the founders were subcloned into pBlueScript (Stratagene) and confirmed by Sanger sequencing.The founders were crossed with WT C57BL/6 J (produced at NUCLEUS, platform for research support of the University of Salamanca) to eliminate possible unwanted off-targets.
LoxP-edited (Mettl1-flox allele) or mice with point mutations/deletions at exon 2 (Mettl1 ± allele) were obtained.Heterozygous mice were re-sequenced and crossed to give rise to edited heterozygous and homozygous mice.Genotyping was performed by PCR with primers F-R and Restriction Fragment Length Polymorphism (RLFP) analysis (1F: 5'-TCT CCT GGT GTG CAT GAA GAC-3'; 1R: 5'-AGG GAA GGT GGT GGA ATC CC -3' .To assess Mettl1 status in established mouse colonies, genotyping was performed by PCR using primers F-5'-ATC TGC AGC CAT AGC CTC AT-3' and R-5'-TCT AAC CTG CTG AGA TGG CA-3' , for floxed alleles, followed by a digestion with Bcl-I restriction enzyme (New England Biolabs) for straight Mettl1 ± mice since CRISPR editing caused a point mutation generating a Bcl-I restriction site.We did not obtain Mettl1-/-mice, suggesting the lethality of the homozygous allele.qPCR confirmed decreased expression of Mettl1 in all tested tissues from Mettl1 ± , or in the prostate of Probasine-Cre x Pten flox/flox x Mettl1 flox/ flox mice.
The Pten flox conditional knockout allele and the prostate epithelium-specific Cre recombinase Pb-Cre4 have been described elsewhere [45] and were provided by Pr P. P. Pandolfi.For prostate tumour evaluation upon Mettl1 deletion, Mettl1 ± or Mettl1 flox/flox mice were crossed to Pten flox/flox and Pb-Cre mice.

Patient samples
RNA and protein samples from prostate specimens were obtained upon informed consent and with evaluation and approval from the corresponding ethics committee (CEIC codes OHEUN11-12 and OHEUN14-14), as described elsewhere [136].

Computational expression and mutation load analysis from human prostate tumour datasets
For expression analyses of human prostate tumours in different disease status or recurrence, we combined the use of CANCERTOOL [137] with complementary ad hoc analyses on various datasets [3,6,7,[42][43][44]138].All downloaded data were pre-processed, background corrected, and log 2 transformed and quantile normalised according to [137].Differential expression was computed by means of Z-scores as the number of standard deviations of expression of cancerous samples relative to the mean of expression in a reference population (primary tumour vs. normal and metastasis vs. normal).Subsequently, an ANOVA test was performed for the different groups.Finally, we calculated coherence as the percentage of datasets in which the gene was significant out of the number of datasets in which the gene appeared (ANOVA P-value significant ANOVA (value < 0.05)).Only genes with consistent differential expression in at least three datasets and 75% coherence were considered significant.Correlation analyses (Pearson's correlations) and data [3,6,7,42,138] were retrieved from CANCER-TOOL [137].
For disease-free survival, expression data were organised into deciles for (TCGA) [3].The Cox proportional hazards model was used to evaluate the association between the survival time of patients and expression levels of primary tumours.For biochemical recurrencefree survival of patient groups from the Cambridge and Stockholm cohorts, data were retrieved from camcAPP [48].
For correlation analysis, data were retrieved from CANCERTOOL [137].

Cloning and generation of stable cell lines
For lentiviral particle production, HEK293FT cells were transfected with calcium phosphate or JetPEI (PolyPlus) using the appropriate expression vector.Cell transfection with lentiviral vectors was performed according to the standard procedures.Briefly, viral supernatants from 48 and 72 h post-transfection were collected and concentrated using a Lenti-X concentrator (Takara).Lentiviral particles were resuspended in 200 μL of complete medium and added to 1 × 10 6 cells in medium supplemented with 8 μg/mL Protamine Sulfate (Sigma).Transduced cells were selected using 2 μg/mL puromycin (Sigma) or 10 μg/mL blasticidin (Santa Cruz Biotechnology) for 3 or 5 days, depending on the selection cassette.
For inducible silencing of human WDR4, two different short-hairping RNA (shRNAs) sequences targeting human WDR4 were cloned into the doxycycline-inducible lentiviral plasmid pLKO.1-TetON(Addgene, 21915).The shRNAs sequences used were: shWDR4-1: GCT GTT GAA ACT CAG CAC GAA; shWDR4-2: GCA CCT GTC TAT GCT GTT AGA and introduced into AgeI and EcoRI restriction sites.A scramble control vector was generated by cloning oligo CCG GCA ACA AGA TGA AGA GCA CCA ACT CGA GTT GGT GCT CTT CAT CTT GTTG.All clones were sequence-validated.Cells were infected with plasmids expressing either scramble control or shRNAs.shRNAs expression was pre-induced for 3 days with 0.1 µg/mL doxycycline (Sigma) and then induced for the experiment length at 0.25 μg/mL or directly induced with 0.5 μg/mL for 3-5 days.The silencing efficiency was tested both by RT-qPCR and western blotting.
HA-METTL1 expressing vector was generated starting from the ORF obtained from Source BioScience (clone: X96698; IMAGE clone: 36006) and cloned into a modified retroviral doxycycline-inducible pTRIPZ vector (Dharmacon) [139] with an HA-tag.Site-directed mutagenesis for the generation of METTL1-AFPA was performed using the Q5 Site-Directed Mutagenesis Kit (NEB) and the primers F: 5'-CCC GCC CCA CAT TTC AAG CGG ACA -3' and R: 5'-GAA GGC GAA GAA CAT CTT TGT CAG C-3' .PC3 parental lines and PC3 METTL1 KO cell lines were transduced with lentiviral constructs containing TRIPZ-HA-METTL1 (wild-type), TRIPZ-HA-METTL1-AFPA, or empty vector.Expression of the tagged version was confirmed by western blotting.Wildtype or mutant METTL1 expression was induced with 0.5 µg/mL doxycycline for 5 days for dot blot analysis and 0.1 µg/ml doxycycline for growth curve assays.

North-dot blotting
Total RNA was extracted from cells in culture or mouse prostate using TRIzol and isopropanol precipitation as previously described.tRNA size selection was performed using the mirVana isolation kit (Thermo Fisher) following the manufacturer's instructions.10 ng of small RNAs (< 200 nt) or large RNAs (> 200 nt) were blotted onto 0.2 μm nitrocellulose membranes (HE Healthcare) using Milliblot D (Millipore).Membranes were crosslinked with 120 mJ/cm 2 in UV Stratalinker 2400 (Stratagene), then blocked with 5% BSA in RNA-free PBS-Tween for 1 h.Anti-m 7 G antibody (MBL, RN017M) was incubated overnight at 4 °C, followed by incubation with anti-mouse HRP-conjugated secondary antibodies at room temperature for 1 h.Signals were detected using homemade ECL.For loading control, membranes were stained with 0.04% methylene blue (Sigma) in 0.3 M Sodium acetate pH 5.2 for 10 min.The methylene blue staining intensity was used to normalize the X-ray films.To measure the m 7 G levels in mouse urine, fresh urine was collected and directly loaded onto nitrocellulose membranes, and the protocol was performed as previously described.Biological replicates are indicated for each experiment.

mRNA expression profiling
For transcriptomic analysis of mouse prostate samples, RNA was extracted from the tissues using a Nucleospin RNA kit (Macherey-Nagel).Four biological replicates were used for genotype.Pten-KO mice were used at 3 and 6 months of age as well as aged-match wild-types.RNA quantity and quality were evaluated using a Qubit RNA HS Assay Kit (Thermo Fisher Scientific, Cat.# Q32855) and Agilent RNA 6000 Nano Chips (Agilent Technologies, Cat. # 5067-1511), respectively.Sequencing libraries were prepared using the "TruSeq Stranded Total RNA Human/Mouse/Rat" Kit (Illumina Inc., Cat. # RS-122-2201), followed by the "TruSeq Stranded Total RNA Sample Prep-guide (Part # 15031048 Rev. E)".Briefly, starting from 500 ng of total RNA, rRNA was depleted and the remaining RNA was purified, fragmented, and primed for cDNA synthesis.The first strand cDNA was synthesised using SuperScript-II Reverse Transcriptase (Thermo Fisher Scientific, Cat.#18064-014).Second-strand cDNA was synthesised using Illumina reagents at 16 °C for 1 h.A-tailing and adaptor ligation were then performed.Finally, library enrichment was achieved using PCR.All the samples were multiplexed and sequenced on a HiSeq4000 platform (Illumina).
For transcriptomic analysis of PC3 WT and METTL1 KO cells, total RNA was extracted from purified cell populations using the NZY Total RNA Isolation kit (RNA Purification, NZYTech) according to the manufacturer's instructions.RNA integrity was assessed using an Agilent 2100 Bioanalyzer (Agilent Technologies, CA).Chip microarray hybridizations and data were generated with Affymetrix Clariom S Assay Human (Affymetrix) (> 20,000 well-annotated genes) were used in this study.
Labelling and hybridization were performed according to Affymetrix protocols.Briefly, 100 ng of total RNA was amplified and labelled using the WT Plus reagent kit (Affymetrix) and then hybridised to the Clariom S human array (Affymetrix).Washing and scanning were performed using the GeneChip System of Affymetrix (GeneChip Hybridization Oven 645, GeneChip Fluidics Station 450, GeneChip Scanner 7G).Raw data were normalised by robust multiarray analysis (RMA) using the oligo package [142] and subsequently analysed to detect genes differentially expressed between PC3 METTL1-KO and WT cells through the limma package [143].Differentially expressed genes in METTL1-KO vs. WT with Benjamini-Hochberg corrected P values < 0.05 were considered significant.

PAR-CLIP analysis
For the analysis of PAR-CLIP form PC3 cells, raw sequence data was demultiplexed based on library specific barcodes, quality filtered (mean sequence quality > 30) and trimmed (removing barcodes, PCR and sequencing adapters) using Flexbar (v.3.5.0) as described in [144].
PAR-CLIP data from the HEK293T cell line expressing Flag-METTL1 were obtained from the GSE100756 [51].The data were downloaded from the UCSC table browser.Reads were aligned to the hg19 genome assembly using bowtie version 1.2.3 with the configuration recommended by PARalyzer developers [145], -v 2 -m 10 -best -strat.The reads were aligned to Repeat Masker sequences after removing the tRNA sequences.Aligned reads were used for peak calling with PARalyzer version 1.5 using the recommended settings.The peaks identified in only one sample were removed.Counts per million associated with the common peaks were used to calculate the coefficient of variation (CV).Only the peaks with a CV lower than 30% were retained.These peaks were annotated according to GENCODE release 19 for further analysis using in-house scripts.

AlkAniline-seq
For m 7 G detection at single-nucleotide resolution, AlkAniline-seq was performed according to [52].Briefly, RNA was prepared from two independent replicates of PC3 WT and PC3 METTL1 KO cells.tRNA fraction (150 ng) was subjected to alkaline hydrolysis in 50 mM bicarbonate buffer pH 9.2 for 5 min at 96 °C.The reaction was stopped by ethanol precipitation using 3 M Na-OAc, pH 5.2 and glycoblue as a carrier in liquid nitrogen.After centrifugation, the pellet was washed with 80% ethanol and resuspended in nuclease-free water.RNA fragments were dephosphorylated using 5 U of Antarctic Phosphatase (NEB, #M0289S) for 1 h at 37 °C.After inactivation of the phosphatase for 5 min at 70 °C, RNA was recovered by a phenol:chloroform:isoamyl alcohol (25:24:1) extraction, followed by a chloroform step and precipitated in 96% ethanol using 3 M Na-OAc, pH 5.2 and glycoblue for at least 2 h at -80 °C.After centrifugation, the pellet was washed with 80% ethanol and resuspended in 20 μL of Aniline 1 M pH 4.5 and incubated in dark for 15 min at 60 °C.The reaction was stopped by ethanol precipitation and incubated overnight at -80 °C.After centrifugation, the pellet was washed twice with 80% ethanol and resuspended in nuclease-free water.RNA fragments were converted to library using NEB-Next ® Small RNA Library kit (NEB, #E7330S) following the manufacturer's instructions.DNA library quality was assessed using a High Sensitivity DNA chip on a Bioanalyzer 2100.Library quantification was done using a fluorometer (Qubit 2.0 fluorometer, Invitrogen, USA).Libraries were multiplexed and subjected for highthroughput sequencing using an Illumina Nextseq2000 instrument with a 50 bp single-end read mode.
Bioinformatic analysis of AlkAnilineSeq data was performed as follows.Raw sequencing reads were trimmed by Trimmomatic v0.39, to remove eventual primer dimers and 3'-adapter sequence.Trimmed reads were aligned by bowtie2 v2.4.2 in end-to-end mode to the optimized human tRNA reference containing 43 representative tRNA species [146] (and positions of reads' 5'-extremities were counted using a custom awk script.Normalized cleavage (NCleavage) was calculated for every RNA species, by dividing of reads' count for a given position to total number of reads mapped to a given RNA [147].

tRNA-seq
For tRNA-seq, tRNA libraries were prepared from PC3 WT and METTL1 KO cells with four independent biological replicates per condition.Total RNA was extracted using TRIzol (Honeywell, 33539), followed by DNAse I Turbo (Invitrogen) treatment.Then, tRNAs (< 200 bp) were selected using the mirVana miRNA isolation kit (Thermo Fisher) following the manufacturer's instructions.tRNAs were de-aminoacylated with a solution containing 1 mM EDTA and 0.1 M Tris-HCl pH 9.0 for 30 min at 37 °C.All small RNAs were treated with T4 PNK (NEB) to ensure the presence of phosphorylated 5' ends and 3'OH ends.tRNA libraries were generated using the NEXTFLEX Small RNA-Seq Kit v3 (Bioo Scientific Corp.) for Illumina Platforms, following the manufacturer instructions.All the samples were multiplexed and sequenced on a HiSeq2500 platform (Illumina).

tRNA-seq analysis
Raw sequencing FASTQ files were trimmed with 'cutadapt' of the 'TGG AAT TCT CGG GTG CCA AGG' adapter, retaining reads with a minimum length of 23 nt.Reads were subsequently trimmed by 4 nt on both ends following the platform recommendations.Reads were aligned to the GRCh38 (hg38) reference genome using bowtie (bowtie-bio.sourceforge.net)with parameters '-m 500 -v 2' , allowing a maximum of 500 multiple mappings and two mismatches.tRNA genomic coordinates for hg38 were obtained from the 'GtRNAdb' database (gtrnadb.ucsc.edu).Reads on tRNA coordinates were processed using 'htseq-count' from the Python package 'HTSeq' (htseq.readthedocs.io)with the '-samout' option, and counts for individual reads representing tRNA fragments and their multiple sequence alignments on tRNAs were extracted from the resulting sam files using custom PERL scripts.Fragments with fewer than 10 counts over all samples were removed, and a pseudo-count of 1 was added.Fragments were filtered according to length: ≥ 18 nt and ≤ 50 nt.Counts of tRNA fragments were normalized, and differential abundance of tRNA fragments was evaluated in 3 replicates per condition (WT or KO) using the R package 'DESeq2' (https:// bioco nduct or.org/ packa ges/ relea se/ bioc/ html/ DESeq2.html).3' tRNAderived fragments (3'-tRFS) were defined as tRFs that end < 10 from tRNA end site, Int-tRFs span the interior of the mature tRNA sequence, 5′tRNA-derived fragments start site > -10 and < 10 of the tRNA start site, from which 5'-halves were > 35 nucleotides long, and all others were < 35 nucleotides and were split into 5'-tRFs (without 5′ terminal oligoguanines) and 5'TOGs (containing a 5′ terminal oligoguanines).

Northern blotting
Total RNA was extracted from cultured cells using TRIzol and isopropanol precipitation, as previously described.To analyse tRNA fragment formation under stress conditions, 200 μM NaAsO 2 was added to the cells for 2 h, then NaAsO 2 was removed and the cells were collected at the indicated time points (h) after NaAsO 2 was added to cells.The 15% polyacrylamide 8 M urea-TBE gels were pre-run for at least 1 h at 180 V in 1 × TBE. 8 to 10 μg of total RNA were loaded onto 15% polyacrylamide gels, run for 1 h at 150 V, and transferred to a Nylon Hybond-NX membrane (GE Healthcare) for 1 h at 45 V by wet electrophoresis (BioRad) in 0.5 × TBE.Membranes were crosslinked with 120 mJ/cm 2 in UV Stratalinker 2400.Membranes were pre-hybridised at 42ºC for at least 1 h in ULTRAhybond buffer (Invitrogen).Single-stranded DNA probes (20 pmol) were radiolabelled at the 5' end by incubating 80 mCi 32 P-ATP (Perkin Elmer) with T4 PNK (NEB) for 1 h at 37ºC.Probes were purified using Oligo Clean and Concentrator (Zymo Research) columns and denatured for 5 min at 95ºC.Membranes were incubated overnight with the radiolabelled probe and then washed twice with low-stringent buffer (0.1 × SSC, 0.1% SDS) at 42 ºC for 15 min and twice with high-stringent buffer (2 × SSC, 0.1% SDS).Membranes were dried and exposed overnight at -80ºC with X-ray films.The following DNA probes were used: 5'-Cys-TAC CCC TGA GCT ATA CCC CC-3' , 5'-Pro-ATA CCC CTA GAC CAA CGA GCC-3' , 5'-Lys-CCG ACT GAG CTA GCC GGG C-3' .Two biological replicates per cell line were analysed.

Flow cytometry
To measure protein synthesis, cultured cells were incubated with 20 μM OP-puro (Medchem Source LLP) for 1 h at 37 ºC in complete medium.Cells were untreated or treated with 200 µM NaAsO 2 for two hours, then NaAsO 2 was removed and cells were collected after 2, 4 or 8 h after NaAsO 2 was added to cells.Cells were collected by trypsinisation and fixed with 1% paraformaldehyde in PBS for 15 min on ice.The cells were washed in PBS and permeabilised with PBS supplemented with 3% fetal bovine serum (Sigma-Aldrich) and 0.1% saponin (Santa Cruz Biotechnology) for 5 min at room temperature.OP-puro conjugation to a fluorochrome was performed by azide-alkyne cycloaddition using the Click-iT Cell Reaction Buffer Kit (Jena Bioscience) and 5 μM of Alexa Cy5.5 conjugated to azide (Jena Bioscience).After 30-min reaction, cells were washed twice in PBS with 3% FBS and 0.1% saponin and then resuspended in PBS.Events were acquired using an Accuri C6 cytometer.Staining and protein synthesis were analysed using FlowJo software (BD Biosciences).
For cell cycle analysis, cells in culture were trypsinised, washed with PBS and medium, fixed in cold 70% ethanol for 10 min, pelleted and resuspended in staining buffer containing 1 µg/mL Propidium Iodide and 50 µg/mL RNase A (Roche) in PBS.The cells were then incubated for 30 min at 4 ºC in the dark and acquired using a BD Accuri C6 flow cytometer (BD Biosciences).Cell debris and aggregates were eliminated by gating the cells forward versus side scatter and FSC-H versus FSC-A.The DNA content was quantified using FlowJo software (BD Biosciences).Three technical replicates were used for each cell type.
For cell death and apoptosis analysis, 70.000 cells were seeded in 12-well plates, after 48 h cells were washed twice with PBS and trypsinised.Cells were pelleted and resuspended in Annexin V Binding Buffer (10 nM Hepes/ NaOH (pH 7,4), 140 mM NaCl 2,5 mM CaCl2) containing 1 mg/mL Propidium Iodide and 10 mg/mL RNase A (Roche). 3 uL of Annexin V (Immunostep, ANXVF-200 T) was added to each sample.Cells were then incubated for 15 min at room temperature in the dark and after that 50 uL of Annexin Binding Buffer was added to each sample.The cells were acquired using a BD Accuri C6 flow cytometer (BD Biosciences).Cell debris and aggregates were eliminated by gating the cells forward versus side scatter and FSC-H versus FSC-A respectively.The percentage of apoptotic cells was quantified using FlowJo software (BD Biosciences).Three technical replicates were used for cell type.

Cap binding assay
m 7 G-cap pulldown was performed as described [57].Briefly, subconfluent WT PC3 cells were collected in lysis buffer (10 mM Tris-HCl, 140 mM KCl, 4 mM MgCl 2 , 1 mM DTT, 1 mM EDTA, 1% NP-40, protease inhibitors and homemade phosphatase inhibitors) and incubated on ice for 30 min.Lysates were cleared by centrifugation at 12,000 rpm for 30 min at 4 ºC and quantified using BCA method (Thermo Fisher Scientific).30 μL of 7-methyl-GTP-Sepharose beads (Jena Bioscience) were washed and incubated with 200 μg of protein extracts diluted in lysis buffer to 1 mg/mL.Incubation was performed overnight with rotation at 4 ºC.The beads were then washed twice with lysis buffer containing 0.5% NP-40 and twice with PBS.Bound proteins were analysed by western blotting.

Proliferation assays
From 3.000 to 15.000 cells were seeded in 12-well plates, with three or six replicates per condition.Cells were maintained in culture medium for up to 8 days, with collection points every two days.Day 0 was defined less than 24 h after seeding.In the case of the catalytic dead mutant proliferation rescue, 0.1 µg/mL of doxycycline was added to the cells during curve seeding.For collection, cells were washed twice with PBS and fixed with 4% paraformaldehyde (Panreac).Staining was performed with 0.1% crystal violet (Sigma) in 10% methanol for 1 h, after which the wells were thoroughly washed with dH 2 O and dried overnight at room temperature.Crystal violet was dissolved in 10% acetic acid for 40 min, and cell density was quantified by measuring the absorbance at 595 nm using a Plate Infinite reader 200 Pro (TECAN).All data were normalised to day 0 measurements.

Cell immunofluorescence of BrdU
Cells were seeded on cover glasses (Thermo Scientific) and grown in complete medium.For BrdU incorporation in S phase, the cells were treated with 10 μM BrdU (Sigma, 19-160) for 1 h.For cell collection, PBS washes were performed, followed by fixation for 10 min with 4% PFA (Panreac Applichem) and permeabilization with 0.2% Triton X-100/PBS (Panreac Applichem) for 10 min.For BrdU staining, coverslips were treated with 2N HCl for 30 min at room temperature and then washed thoroughly.Blocking was performed for 1 h at room temperature with 5% goat serum (Gibco) in PBS.Primary antibodies diluted in blocking solution were incubated overnight at 4 ºC, using the following primary antibodies: anti-BrdU (Abcam, ab6326).The next day, the cover glasses were washed three times with 0.05% Tween-20 in PBS and incubated for 1 h with fluorophore-conjugated secondary antibodies at room temperature in the dark.Donkey/goat anti-mouse or anti-rabbit secondary antibodies conjugated to Alexa Fluor 488 or Alexa Fluor 594 were used.Nuclei were stained with 0.5 μg/ mL of DAPI (Sigma, D9542) for 10 min, and cover glasses were mounted with Mowiol (Sigma, 81381).Fluorescence images were acquired with a Leica SP5 confocal microscope.

Proteomic analysis
For total protein analysis, a sub-confluent p100 plate was collected per condition.Cells were trypsinised and washed three times with PBS, and cell pellets were snapfrozen in liquid nitrogen.The cells were then lysed in buffer containing 2 M thiourea, 7 M Urea, 4% Chaps and 200 mM DTT.The extracts were cleared by centrifugation.For actively translated proteins, 24 × 10 6 cells were treated with 20 μM OP-puro (Medchem Source LLP) for 1 h at 37 ºC in complete medium and snap-frozen in liquid nitrogen after collection.The alkyne-tagged proteins were captured in an azide-tagged resin by a Cu(I)catalysed azide-alkyne cycloaddition reaction, which was performed using a Click Chemistry Capture Kit (Jena Bioscience).Bound proteins were then washed with high-stringency buffer and detached from the resin upon protease digestion.Mass spectrometry was performed to analyse protein composition.
For in-solution digestion, the proteins were extracted using 7 M urea, 2 M thiourea, and 4% CHAPS.Samples were incubated for 30 min at RT under agitation and digested according to the FASP protocol described previously [148] with minor modifications.Trypsin was added at a trypsin:protein ratio of 1:10, and the mixture was incubated overnight at 37 °C.For on-bead, the proteins were denatured in 8 M urea, reduced with DTT (5 mM), and alkylated with iodoacetamide (25 mM).The sample was diluted in ammonium bicarbonate to 1.5 M urea, and trypsin was added to the same ratio described above and incubated overnight at 37 °C as described elsewhere.The resulting peptides in either strategy were concentrated in a speed-vac and desalted using C18 stage tips (Millipore) prior to acquisition.For mass spectrometry analysis, the samples were analysed using a hybrid trapped ion mobility spectrometry -quadrupole time of flight mass spectrometer (timsTOF Pro with PASEF, Bruker Daltonics) coupled online either to a nano Elute (Bruker) or EvoSep ONE liquid chromatograph (EvoSep).Protein identification and quantification were carried out using MaxQuant software [149] with default settings, except for LFQ min ratio count of 1. Searches were performed against a database consisting of human protein entries (Uniprot/Swissprot) with precursor and fragment tolerances of 20 ppm and 0.05 Da respectively.Only proteins identified with at least two peptides with an FDR < 1% were considered for further analysis.Data (LFQ intensities) were loaded onto Perseus platform [150] and further processed (log 2 transformation and imputation).
The Gene Ontology web tool powered by Panther (http:// geneo ntolo gy.org/) was used for functional annotation analysis of differentially expressed gene sets to identify specific gene subsets sharing co-occurrent functional annotations linking them, with high statistical significance, to particular Gene Ontology (GO) Biological Process categories.Fisher's exact test was applied with Bonferroni correction.

Polysome profiling
Six 150 cm plates were seeded per replicate, and 80% confluent plates were treated for 5 min at 37 °C with 0.1 mg/ mL of cycloheximide (CHX, Santa Cruz Biotechnology).Plates were washed twice with PBS supplemented with 0.1 mg/mL CHX and then collected with lysis buffer (20 mM Tris-HCl, 100 mM KCl, 5 mM MgCl 2 , 0.3% Igepal, 0.1 mg/mL CHX, protease inhibitors).RNA-protein content was measured using a Smart-Spect Plus spectrophotometer (Bio-Rad) and equal amounts of extracts were loaded into the gradients.7-50% sucrose gradients (20 mM Tris-HCl pH7.5, 100 mM KCl, 5 mM MgCl 2 , 1% CHX) were previously prepared and balanced, and translation components were separated by ultracentrifugation for 2:45 h at 4 ºC and 39,000 rpm with no break.Gradients were fractionated using a Density Gradient Fractionation system (Brandel) and 500 μL fractions were collected.Half the volume of each fraction was used for RNA extraction with warm phenol protocol.Briefly, equal volumes of the fractions were mixed with 60 ºCheated Phenol-Chloroform and a solution containing 10 mM Tris, 350 mM NaCl, 10 mM EDTA, 1% SDS and 42% urea (Fisher Chemical), followed by RNA precipitation with isopropanol.cDNA and RT-qPCR were performed on 16 different fractions from each replicate.To calculate the content of mRNA in each separated fraction, the mRNA content of each transcript in each fraction was normalized with the content of the transcript in all the analysed fractions and represented as the proportion or fraction of the total.

Cytokine array
For human cells, on day 0, PC3 WT and METTL1 KO cells were seeded in 6-well plates.On day 3, the medium was removed and 1 ml of serum-free DMEM was added to each well.On day 5, the medium was collected and centrifuged again.The supernatant was used to analyse the expression levels of multiple proteins using the RayBio C-series Human Inflammation Antibody Array 3 Kit (RayBiotech, AAH-INF-3-2).For mouse tissues, prostate tissue from five-months old Pten-KO/ Mettl1 + / + and Pten-KO/Mettl1 fl/fl mice were used.Protein extracts from three mice of each genotype were used to analyse multiple protein expression levels using Mouse Cytokine Array G3 kit (RayBiotech, AAM-CYT-G3-8).

Phagocytosis assay
The cells were treated in the same manner as for the polarisation assay, but this time 10.000 THP-1 were seeded in 24-well plates.On day 5, the medium was replaced with HBSS medium supplemented with 2% FBS and 50 ug of pHrodo ™ Red E. coli BioParticles ™ Conjugate for Phagocytosis (Thermo Fisher) was added to each well.Fluorescence intensity was measured for 4 h taking pictures every 5 min with an Eclipse TE2000 Inverted Microscope (Nikon).

Monocyte isolation from peripheral blood and macrophage polarisation
PBMC were isolated from blood samples by centrifugation using the density gradient medium Lymphoprep ™ (Stem Cell Technologies).Monocytes were isolated from PBMC using anti-human DynabeadsTM CD14+ (Thermo Fisher), according to the manufacturer's instructions.500.000monocytes were seeded in 6-well plates at day 0, adding 5 ng/mL Granulocyte-Macrophage Colony-Stimulating Factor (GM-CSF) was added in order to differentiate monocytes to Mø.On day 3, conditioned media from PC3 WT and METTL1 KO cells was added.As controls, on day 6, some cells were polarised to M1 by adding 100 ng/mL LPS and 20 ng/ mL IFNγ, and others to M2 by adding 20 ng/mL IL-4 and 20 ng/mL IL-13; M2 controls were seeded on day 0 with 50 ng/ml (Macrophage Colony-Stimulating Factor (M-CSF) instead of GM-CSF.Some cells were cultured for 6 days with medium and 5 ng/mL GM-CSF as the Mø control.On day 7, cells were detached with PBS containing 2.5% FBS (Gibco) and 0.5 Mm EDTA for 30 min, and polarisation was analysed using a BD FACSAria III flow cytometer (BD Biosciences).Antibodies against anti-CD11b (BioLegend, 101206), anti-CD14 (BD-Horizon, 562335), and anti-CD68 (BioLegend, 333813) were used.Cell debris and aggregates were eliminated by gating the cells forward versus side scatter and FSC-H versus FSC-A respectively.Polarisation was analysed using FlowJo software (BD Bioscience).Three technical replicates were used for condition.

T cell isolation from peripheral blood, proliferation and migration assays
PBMC were isolated from blood samples by centrifugation using the density gradient medium Lymphoprep ™ (StemCell Technologies).PBMC cells were cultured on day 0 in a 75 mm 3 flask adding 100U/mL IL-2 (Miltenyi Biotec) every 2 days.On day 7, cells were collected and stained with the Cell Trace CFSE Cell Proliferation Kit (Thermo Fisher), as indicated by the manufacturer.
For proliferation measurements, lymphocytes were then co-cultured with 500.000cells/well to previously polarised macrophages following the macrophage polarisation protocol described above.On day 9, the supernatant containing the lymphocytes was collected and proliferation was analysed using a BD FAC-SAria III flow cytometer (BD Biosciences).Antibody anti-CD3 (BioLegend, 300428) was used to select T lymphocytes.
For migration assays, peripheral blood monocytes were seeded in 24-well plates and polarised with PC3 WT and METTL1 KO cells conditioned or with specific cytokines as previously described.On day four of seeding, TC-inserts with a pore size of 8 μm (Sarstedt, 83.3932.800)were placed in each well.The same number of lymphocytes previously stained with CFSE was seeded in the upper compartment.After three hours, the medium from the lower compartment was collected, the volume was analysed using a BD FACSAria III flow cytometer (BD Biosciences), and the number of CFSE-positive cells was used as an indicator of lymphocyte migration.Monocytes and lymphocytes were isolated from the same healthy donors in each experiment.

Analysis of abundance of cell types in human tumours
CIBERSORTx [151] and TIMER2.0[89] were used to assess cellular abundance and cell type-specific gene expression patterns from bulk tissue transcriptome data obtained from The Cancer Genome Atlas Prostate Adenocarcinoma (TCGA-PRAD) collection [152] using cBio-Portal [153,154] and Taylor datasets [6].For correlation analyses, only tumours with high METTL1 expression were analysed.For TCGA samples, METTL1 > 500 normalised RPMs.For Taylor samples, METTL1 > 6.85 log 2 normalised expression.

Anti-PD1/CTLA-4 and anti-CD8 mouse treatments
At the age of four months, a combination of anti-mouse PD1 (Bio X Cell, BE0146) and anti-mouse CTLA-4 (Bio X Cell, BE0131) was administered at a dose of 5 μg of each antibody per gram of mouse, both antibodies in 100 μL.To eliminate CD8 T cells, anti-mouse CD8α (Bio X Cell, BE0061) was administered at a dose of 100 μg per mouse.As control treatment, a combination of rat IgG2a isotype control (Bio X Cell, BE0089) and polyclonal Syrian hamster IgG (Bio X Cell, BE0087) was used as a control at the same dose.All treatments were administered by intraperitoneal injection every three days for eighteen days.Mice were weighed every three days to discard treatment toxicity.The in vivo response to immune checkpoint blockade or CD8 T cells inhibition was evaluated by measuring each prostate lobe volume one week after the treatment regime was completed.

Statistical analysis
Statistical analyses were performed using GraphPad Prism 8.2.P values less than 0.05 were considered statistically significant.For in vitro experiments, continuous and normal distribution was considered; applying unpaired one (validation or hypothesis-driven experiment) or two (experimental design without predicted result) tailed Student's t-test, whenever indicated.In vivo experiments with ten or less replicates were considered normally distributed, and a parametric Student's t-test was applied.For more than ten replicates, unpaired Student's t-test was applied if the distribution was normal and Mann-Whitney test if the distribution was not normal.For the comparison of multiple samples, a oneway or two-way ANOVA test was applied.Data are represented as mean ± s.d. or s.e.m. when indicated, mean with interquartile range or boxplot with line at mean or min and max.Pearson's correlation was computed for normally distributed samples and Spearman's correlation for non-normally distributed samples."n" represents the number of independent experiments, individual mice or patient specimens.At least three technical replicates were used.No statistical analysis was performed to determine the sample size.The experiments were randomized.The investigators were not blinded to allocation during the experiments and outcome assessment.

Data access
The transcriptomic data, PAR-CLIP and microarray generated in this publication have been deposited in NCBI's Gene Expression Omnibus (GEO Series accession numbers GSE189883, GSE190911, and GSE230694) and the European Nucleotide Archive (ENA projects PRJEB49285 and PRJEB53441).The mass spectrometry proteomics data were deposited in the ProteomeXchange Consortium via the PRIDE partner repository with the dataset identifier PXD030141.

Fig. 2 (
Fig. 2 PI3K-AKT-mTORC pathway mediate upregulation of METTL1 expression in PCa.A Correlation analysis between METTL1 and PTEN expression in human primary prostate tumours expression datasets.Plotted values correspond to the log 2 -normalised gene expression values for each patient in the indicated dataset.The black line represents linear regression, grey area indicates the limits of the confidence intervals.Pearson's correlation coefficient (R) and p-values are indicated.Grasso n = 88; Taylor n = 183; TCGA n = 497.B, C METTL1 expression is regulated downstream of AKT signalling.Western blot (B) and RT-qPCR (C) analyses of METTL1 expression upon PI3K pathway inhibition in DU145 cells.DMSO as vehicle (Veh), BKM-120 (BKM) as pan-PI3K inhibitor, MK2206 (MK) as AKT inhibitor, rapamycin (RAPA) as mTORC1 inhibitor, and Torin (TOR) as mTORC1/2 inhibitor.For western blotting, cells were treated for 48 h, and for RT-qPCR, for 8 h.Means ± SD are shown (n = 2) (B) and (n = 6) (C).D Stratification of patients with a worse prognosis according to high METTL1 levels and low PTEN expression.Kaplan-Meier curves representing the disease-free survival (DFS) of patient groups selected according to Q1 (Pten L ) and Q4 (Pten H ) quartile expression of PTEN, and METTL1 high (Met H , log 2 -normalised expression > 8.72) and METTL1 low (Met L , log 2 -normalised expression < 8.72) in recurrent and disease-free (DF) tumour samples from the TCGA dataset.E-G Mettl1 is highly expressed in mouse prostate tumours.Mettl1 expression analysis in Pten-KO mice at 3 and 6 months of age compared to wild-type mice (WT) at the same ages by western blot (E, F) and RT-qPCR (G).* in (E) indicates an unspecific band.Means ± SD are shown for three (F) and five replicates (G).H, J Increased 7-guanine tRNA methylation in RNA from prostate tumours compared with that in normal prostate.North-dot blot of m 7 G levels in long RNAs (> 200 nucleotides long RNAs) and tRNAs extracted from prostatic tissue of Pten-KO mice with intraepithelial prostatic neoplasia (at 3 months of age) or invasive tumour (at 6 months), and wild-type (WT) mice of the same age.Methylene blue staining was used as the loading control (H, bottom panel).Means ± SD are represented (n = 4) (J).I, K Increased 7-guanine tRNA methylation in the urine of mice bearing PCa tumours.North-dot blot of m. 7 G levels from the urines of Pten-KO mice with invasive tumour (at 6 months), and wild-type (WT) mice of the same age.Means ± SD are shown (n > 3).L Representative images of immunostained sections for Mettl1 and markers for luminal (AR), and basal (K14) cells in Pten-KO prostates (anterior lobes) at initiation (3 m) and in invasive carcinoma (6 m) and in age-matched wild-type (WT) prostates.Scale bars represent 50 μm.Statistical tests: one-tailed Student's t-test (C), log-rank test (D), Mann-Whitney test (F, G, J, K). *p < 0.05, **p < 0.01, ***p < 0.001 (See figure on next page.)

Fig. 3
Fig. 3 METTL1 preferentially methylates tRNAs.A tRNAs are the most common RNA species bound to METTL1.Upper panel: Boxplot representing the median Log 2 fold change of reads per million (RPM) per transcript bound to METTL1 in PC3 cells.Lower panel: The bar plot shows the total number of unique genes bound to METTL1.B Validation of the lack of METTL1 expression and m 7 G tRNA methylation in PC3 METTL1 KO cell lines.Western blot of METTL1 (upper panel) and north-dot blot of m 7 G levels (lower panel).Proteins, long RNAs (> 200 nucleotides), and tRNAs (< 200 nucleotides) were extracted from three independent PC3 METTL1 KO clones and three control clones.Parental PC3 cells are shown.C LC-MS analysis of m 7 G levels in tRNAs isolated from PC3 METTL1 KO, control, and parental PC3 cells validate the absence of m 7 G in tRNAs extracted from METTL1 KO cells.Means ± SD are represented (n = 3).D Normalised cleavage signals for the tRNAs AlaAGC and PheGAA in PC3 WT and METTL1 KO cells.Letters on the right show the bases where methylation occurs.E Heatmap showing normalised cleavage values for all tRNAs with guanosines at position 46 in PC3 WT and METTL1 KO cells (n = 2 for each genotype).F tRNA secondary structure showing METTL1-methylated guanines (red circles) in the variable loop.G Graphical summary of tRNA isoacceptors methylated by METTL1 (red) and non-methylated or non-transcribed (grey).Statistical tests: two-tailed Student's t-test (C).****p < 0.0001

Fig. 4
Fig.4 The lack of m7 G methylation in tRNAs leads to 5'tRNA fragment accumulation.A Heatmap showing no differences in the levels of mature tRNAs (log 2 normalised reads; RPKM) measured using tRNA-seq data.Rows represent individual tRNA isoacceptors and columns represent independent replicates.tRNA isoacceptor expression levels are row-scaled (each row is normalised to their mean expression and standard variation).B The top pie chart represents the percentage of all abundant tRNA fragments identified in PC3 METTL1 KO cells compared to that in WT cells.The bottom pie chart represents the percentage of differentially express tRNA fragments: fragments with log 2 fold change (FC) > 2 and p value < 0.05 in PC3 METTL1 KO cells versus WT cells.The pie charts show that while 5'tRNA fragments are very abundant in KO cells, 5'TOGs are significantly overexpressed in KO cells compared to WT cells.3'tRFS: 3' tRNA fragments; Int-tRFs: internal tRFs; 5'tRNA: 5'tRNA fragments > 18 and < 35 nucleotides; 5'-halves: 5'tRNA fragments > 35 nucleotides; 5'TOGs: 5'tRNA fragments > 18 and < 35 nucleotides, with 5′ terminal oligoguanine.C The boxplot shows the log 2 fold change (FC) of all fragments (not just the differentially expressed tRNA fragments) in PC3 METTL1 KO versus WT cells.D Size and abundance (density) of 5'TOGs Cys-derived (5G) or Ala-derived (4G) fragments differentially expressed (log 2 FC > 1.5, > 18 nt, and p < 0.05) in PC3 METTL1 KO cells versus WT cells.E Summary of 5'TOGs formed in PC3 METTL1 KO cells.F Increased tRNA fragmentation was observed by northern blot detection of Cys-derived 5'TOGs in PC3 METTL1 KO vs. WT cells (2 technical replicates of 2 biological replicates are represented) (left panel).The boxplot of the right shows quantification of densities of Cys-derived 5'TOGs formed versus full-length tRNAs.G, H Cys-derived 5'TOG fragments are induced by stress.Northern blot detection of Cys-derived 5'TOG in PC3 and DU145 METTL1 KO and WT cells unexposed (0 h), or after 2 and 8 h of oxidative stress exposure.The boxplots of the right show the fraction of Cys-derived 5'tRFs formed normalised to the fraction of full length tRNAs.Values are from two biological replicates, shown in this figure and in Supplementary Fig. S4G and S4J.The loading control of tRNA is shown in the bottom panel as red-safe staining (F-H).Bands corresponding to full length tRNAs are indicated with stars and 5'tRNA fragments are indicated with arrows.Statistical tests: one-tailed Student's t-test.**p < 0.01 (F-H)

Fig. 5
Fig. 5 METTL1 downregulation suppresses protein synthesis, proliferation and tumour growth in vivo.A Global protein synthesis rate measured by flow cytometry analysis of OP-puromycin (OP-puro) incorporation reflects reduced protein synthesis in PC3 METTL1 KO cells compared to the control (WT).Fluorescence was normalised to cell size (FSC) in WT and METTL1 KO cells.Two biological and three technical replicates and the mean ± SD are shown.B Translation initiation and regulatory factors are displaced from the cap of mRNAs in METTL1 KO cells.Log 2 fold change (FC) binding of the indicated translation initiation and regulatory factors to m 7 G-cap-coated sepharose beads in PC3 METTL1 KO vs. WT cells.Densitometry data were normalised to the input.Mean ± SEM, n = 3. C Anti-TOG RNAs block the 5'TOG-dependent displacement of translation initiation factors in vitro.Log 2 FC of translation initiation factors bound to synthetic biotinylated-5'TOG in PC3 WT cells transfected with 5'TOG + Anti-TOGs (ANT) compared to PC3 WT cells transfected with 5'TOG + scramble RNAs (TOG).Densitometry data were normalised to the input.Mean ± SEM, n = 4. D Displacement of translation initiation factors from mRNA caps is TOG-dependent and can be reversed by expressing anti-TOG RNAs.Log 2 fold change (FC) of m 7 G-cap-bound translation initiation factors in PC3 WT and METTL1 KO cells transfected with biotinylated-5'TOG (TOG) or anti-TOG RNA (ANT) versus cells transfected with scramble RNA oligonucleotides.Densitometry data were normalised to the input.Mean ± SEM, n = 3. Original wester blots are shown in supplementary figure S5 (B, C, D).E Growth curves of PC3 METTL1 KO, WT, and parental cells (PC3).Mean ± SD, n = 3.The dotted line represents the average growth of WT and parental or KO cells.F Reduced cell division rate, as measured by BrdU incorporation.G METTL1 depletion increased apoptosis in PC3 METTL1 KO cells.Flow cytometry analysis of Annexin V staining.Mean ± SD, n = 3. H Reduced spheroid formation capacity in PC3 METTL1 KO cells.Means ± SD, n = 3.The dotted line represents the average values of all WT or KO cells.I Tumour growth in xenografted PC3 METTL1 KO and WT cells in athymic nude mice reflects impaired tumour formation in the absence of METTL1.Mean ± SEM, n = 10.J-L Protein expression (J) and m 7 G methylation levels of tRNAs (K, L) of PC3 METTL1 KO cells ectopically expressing a doxycycline-inducible HA-tagged wild-type (WT) or a catalytic dead mutant (AFPA) version of METTL1.PC3 METTL1 KO cells were infected with an empty vector (eV) as a control.Methylene blue staining was used as the loading control (L, bottom panel).Mean ± SD, n = 3 (L).M, N Proliferation (M) and spheroid formation capacity (N) were dependent on METTL1 catalytic activity.PC3 METTL1 KO cells re-expressing METTL1 (WT) or catalytic dead mutant (AFPA) compared to METTL1 KO cells infected with empty vector (eV).Mean ± SD, n = 6.O, P 5'TOG transfection induces apoptosis and reduces cell proliferation.Percentage of apoptotic (O) and growth rates (P) of PC3 METTL1 KO and WT cells after transfection with synthetic 5'TOGs (TOG) or anti-TOG (ANT) RNAs.Controls (Cont) were transfected with scramble RNAs.Mean ± SD, n = 6 (O), n ≥ 10 (P).Statistical tests: Two-way ANOVA (E, I, L), one-way ANOVA (F, G, H), and one-tailed Student's t-test (A-D, M-P).*p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001

D
Fig.6 Loss of m7 G tRNA methylation results in distinct translational programmes.A Schematic overview of nascent polypeptide OP-puro-labelling and enrichment followed by LC-MS/MS peptide identification and quantification analysis (upper panel).Lower panel shows common differentially expressed nascent proteins ranked in a volcano plot according to their statistical p-value (-Log 10 pV) and their relative abundance ratio (Log 2 FC) between four replicates of PC3 WT and METTL1 KO cells.Coloured dots represent statistically (p-value < 0.05) upregulated (red) and downregulated (blue) proteins in METTL1 KO cells.B Gene Ontology (GO) category enrichment of biological processes in significantly (p-value < 0.05) upregulated (UP, FC > 1) or downregulated (Down, FC < 1) nascent proteins in METTL1 KO cells compared to WT cells.Categories were ranked according to their statistical P-value (-Log 10 pV) and fold enrichment of genes found for each category.C No correlation between differentially expressed proteins (protein log 2 FC) and their mRNA (protein log 2 FC) was observed in PC3 WT vs. METTL1 KO cells.Coloured dots represent significant (p-value < 0.05) differentially expressed proteins (blue), mRNAs (yellow) or both (red) for each gene.D Representative polysome profile in WT (grey line) and METTL1 KO PC3 cells (red line) (left panel) shows reduced translation in METTL1 KO cells.The fraction of the abundance of each mRNA in each polysome fraction is shown with respect to the content in all fractions, reflecting increased translation of IRF9 and ISG15 in METTL1 KO cells.Mean ± SEM, n = 3.The right boxplot represents the fold change (FC) of mRNA content in the polysome fraction relative to non-polysome fractions.E Increased IRF9, ISG15 and STAT1 protein expression observed in PC3 METTL1 KO cells is 5'TOG-dependent.Protein expression levels are represented as log 2 fold change for WT (grey bars) transfected with 5'TOG RNA (TOG) versus scramble control RNA (Ct), and METTL1 KO cells (red bars) transfected with anti-TOG RNAs (ANT) versus scramble control RNAs (Ct).Mean ± SEM, n = 3. Original western blots are shown in supplementary figure S7.F, G STAT1-dependence of apoptosis (F) and growth rates (G) of METTL1 KO and WT PC3 cells.KO cells were transfected with siScramble or siSTAT1.Mean ± SD, n = 6 (F, G).Statistical tests: One-tailed Student's t-test was used (E-G).*p < 0.05, **p < 0.01, ***p < 0.001 (See figure on next page.)

(
See figure on next page.)Fig. 7 METTL1 low expression in PCa is associated with increased cytotoxic infiltration and good response to ICB treatment.A Cytokine content in PC3 METTL1 KO compared to WT cell-conditioned media shows upregulation of pro-inflammatory (red) and downregulation of anti-inflammatory (blue) cytokines in METTL1 KO cells.The mean ± SD log 2 fold change is shown (n = 4).The full expression array is shown in supplementary figure S8.B The upper panel shows a schematic overview of the workflow followed to analyse M1-or M2-like endotype polarisation of THP-1-derived Mø macrophages exposed to METTL1 KO and WT cell-conditioned medium (c.m.).The lower panel shows the T-distributed stochastic neighbour embedding (tSNE) analysis of macrophage polarisation (n = 3).C Proliferation of human peripheral blood CD3+ T cells co-cultured with primary Mø macrophages exposed to PC3 WT or METTL1 KO cells' c.m. (n = three technical, two biological replicates).D Migration of human peripheral blood CD3+ T cells towards primary macrophages exposed to PC3 WT or METTL1 KO cells' c.m. (n = three technical, two biological replicates).E Correlation between METTL1 expression and the immune cell infiltrates of M1-like macrophages (CD86), M2-like macrophages (CD163), and CD8+ T cells in human PCa samples.Immunostainings are shown in supplementary figure S8.F Prostate tumour volume from Pten-KO/ Mettl1 + / + and Pten-KO/Mettl1 flox/flox five-month-old mice reflects reduced tumour burden after conditional Mettl1 deletion.Ventral (V), dorsal (D) and anterior (A) lobes.Mean ± SD, n ≥ 5. G Conditional deletion of Mettl1 resulted in reduced tumour proliferation (Ki67 + cells) and increased immune infiltration of iNOS + (M1-like) macrophages and CD8+ T cells.Staining of tumours from Pten-KO/Mettl1 + / + (+ / +) and Pten-KO/Mettl1 flox/ flox (fl/fl) five-month-old mice.Mean ± SD, n ≥ 5, > 10 images per biological replicate.H Fold change of cytokines content in Pten-KO/Mettl1 flox/flox versus Pten-KO/Mettl1 + / + tumours (n = 3).I Significant decrease in tumour volume (fold change: FC) in Pten-KO/Mettl1 flox/flox (fl/fl) mice treated with anti-PD1 + anti-CTLA4 antibodies compared to untreated controls (IgG).Pten-KO/Mettl1 + / + (+ / +) mice tumour volume did not change after anti-PD1 + anti-CTLA4 treatment.Mean ± SD, n ≥ 6. J METTL1 mRNA expression levels in anti-PD1 responders and non-responders in clinical trials of breast cancer, ovarian cancer, colorectal cancer, and glioblastoma (n = 484).The data were retrieved from the ROC plotter.Statistical tests: Pearson's correlation (r), p-value (pV), and linear regression with 95% confidence (bands) are shown (E).One-tailed Student's t-test (A, C, D, G), Mann-Whitney test (F, I, J), *p < 0.05, **p < 0.01, ***p < 0.001 intron sequence.An 872nt single-strand DNA (METTL1 ssODN: CGG GTA AAT AAA AAT TTT AAA ATA TAC ATA AAC AAA GTA GTG CTC CAG TGC CTT AGG CAG TCA GTC TCT GGG AAA ACC TTA CAC ACA TCA GTA CCC AGA AGA ACT TGG CAG ATC TGC AGC CAT AGC CTC ATG ATG GGA CTG ATT CAC ACC CTA GAA GCC AAT AAC TTC GTA TAA TGT ATG CTA TAC GAA GTT ATG GGG ACT TCC CAT TAC ACC TTC TAC ATA CCA CAT GGT CCC TGC ATG CAG AGT TTT TTT CTA AGC CTA CCA CCC CAC CCC CAA CTC TTA CAC AAA ATG TCT GCA CTG GTA TGT ACA ACA GTC TCA TGT GTC TCA TGT CTT CTT TCT AAT AGC CCT GTG AAG CCA GAG GAA ATG GAC TGG TCT GAG CTT TAC CCA GAG TTC TTT GCT CCG CTT ATT CAA AAT AAG AGC CAT GAT GAT CCA AAA GAT GAG AAA GAA AAG CAC TCT GGG GCC CAA GTG GAG TTT GCA GAC ATA GGC TGT GGC TAT GGT GGC TTG TTA GGT AAT AAG CTC GCC CTT TTC TTG GGA CAG GGA GAG GCC TGG GTT CTG CCA TCT CAG CAG GTT AGA GGC AGG ATT AGT TGA CCT TTC CTC CTG GGA CCA GAC AGC AGT GAC ATC AGT GTG GAG AGC CTC CAC CTT CCT TCT ACT CTG GGT CAT AAC TTC GTA TAA TGT ATG CTA TAC GAA GTT ATT GTG GCC TTT GCC CTG GAG GAG GGA GGG GGC TGC CTC AGA TGT ATC TGA GAA CCC CGG TTG TTC TGT CTA GCT CCA GTG GCT CCA GTC CCT CCG TGA GCC ATT CCA CTG CCT TGT GCT GGG TCA GTG AGC CGG GTC AGC ACA AGC GTC AGA GAT CAG CTC CAG TGG CCA CTG ATC CTG AA) was designed as a template for homologous recombination.The METTL1 ssODN donor DNA sequence contained two LoxP sequences in the same orientation flanking exon 2 and two single mutations destroying the PAM sequences.

5 'Additional file 7 :
TOGs bind to translation initiation factors.A) No differences in protein expression of translation initiation complex factors were detected by western blotting in PC3 parental, WT, and METTL1 KO cells.The bottom boxplot shows the normalised protein level quantification, and the dotted lines indicate the mean protein levels in WT cells.B) Global protein synthesis rate measured by OP-puromycin (OP-puro) incorporation in PC3 METTL1 KO cells compared to the control (WT) after NaAsO 2 exposure at the indicated time points.Fluorescence was normalised to cell size (FSC).n = 3; mean ± SD.C) Representative western blot showing translation initiation factors pulled down by m 7 G-cap sepharose beads in PC3 METTL1 KO vs. WT cell extracts.D) Representative western blot showing translation initiation factors pulled with synthetic biotinylated-5'TOG in PC3 WT cells transfected with biotinylated-5'TOGs (5'TOG), anti-TOG (ANT) or scramble RNA (-).E) Representative western blot showing mRNA-cap-bound translation initiation factors after 5'TOG (TOG), anti-TOG (ANT), or scramble control (Ctrl) transfection in PC3 WT and METTL1 KO cells.Quantification of western blots are shown in Fig. 5 (C, D, E).Statistical tests: One-tailed Student's t-test (A), and multiple t-test (B); ***p < 0.001, Additional file 6: Supplementary Figure S6.METTL1 mediated methylation promotes cell proliferation, and tumour growth in vivo.A) Proliferation of DU145 METTL1 KO and WT cell clones.Mean ± SD, n = 6.The thick dotted line represents the average growth of the WT and KO cells.B) Proliferation of METTL1-silenced 22Rv1 cells.Mean ± SD, n = 3. C) Cell cycle analysis of PC3 METTL1 KO, WT, and parental cells showing decreased cell cycle progression in METTL1 KO cells.Mean ± SD, n = 3.The dotted lines indicate the mean of three biological replicates.D) Representative images of BrdU staining of three clones of PC3 METTL1 KO and WT cells.Quantification is shown in Fig. 5. E) Schematic representation of generation of single-cell-derived spheroids, and representative images of singlecell-derived spheroids (lower panel).Quantification is shown in Fig. 5. F) METTL1 depletion affects tumour growth in vivo.Immunohistochemical staining for METTL1, Ki67, and cleaved caspase 3 (Cl-Casp3) in PC3 METTL1 KO and WT cell-derived xenografts.Right charts show the quantification of Ki67 + and Cl-Casp3 + cells per microscopic field.Mean ± SD, n = 5, and ten images per biological replicate.G-I) Protein expression (G) and m 7 G methylation levels (H, I) of a second PC3 METTL1 KO cell clone (KO2) ectopically expressing an HA-tagged wild-type (WT) or catalytic dead mutant (AFPA) version of METTL1.METTL1 KO2 cells were infected with the empty vector (eV) as a control.Methylene blue staining was used as the loading control (H, bottom panel).Mean ± SD, n = 3 (H).J, K) Proliferation (J) and spheroid formation capacity (K) of PC3 METTL1 KO2 cells re-expressing METTL1 (WT) or catalytic dead mutant (AFPA) compared with METTL1 KO2 (eV).Mean ± SD, n = 6.L) WDR4 mRNA expression and cell growth of PC3 cells expressing doxycycline-inducible SCR shRNA or two shRNA against WDR4 with or without doxycycline induction.Mean ± SD are represented (n = 6).M) WDR4 mRNA expression and tumour growth of xenografts of cells expressing doxycycline-inducible SCR shRNA or a shRNA against WDR4 with or without doxycycline induction.Mean ± SEM are represented (n = 10 for shWD2, n = 5 for SCR, for each Dox condition).Scale bar represents 100 μm (D), 50 μm (F).Statistical tests: Two-way ANOVA (A, I, J), oneway ANOVA (C), and one-tailed Student's t-test (B, F, K, L, M). **p < 0.01, ***p < 0.001, ****p < 0.0001.Supplementary Figure S7.Translational and transcriptional characterisation of METTL1 KO cells.A) Differential mRNA expression ranked in a volcano plot according to their statistical P-value (-Log 10 pV) and their relative abundance ratio (Log 2 FC) between PC3 WT and METTL1 KO cells.Coloured dots represent statistically significant (p < 0.05) upregulated (red) and downregulated (blue) mRNAs levels.n = 3. B) GAPDH and KIF20A were not differentially translated in PC3 METTL1 KO cells.Percentage per fraction of mRNA abundance of the indicated genes found in polysomes of WT and METTL1 KO cell lysates.n = 3, mean ± SEM.The boxplot shows the fold change (FC) of mRNA abundance in polysome fractions of METTL1 KO versus WT cells.C) Global protein expression differences ranked in a volcano plot of PC3 METTL1 KO versus WT cells.D) Gene Ontology (GO) category enrichment of biological processes in significantly (p < 0.05) upregulated (UP) or downregulated (Down) proteins identified in PC3 METTL1 KO cells compared with WT cells.E) Western blots of PC3 WT and METTL1 KO cells transfected with synthetic 5'TOG RNAs (TOG), synthetic anti-TOG RNAs (ANT), or scramble RNAs (Ct).High (h) and low (l) western blot exposures are shown for pSTAT1, total STAT1 and IRF9.The bottom bar plots indicate the expression of each protein relative to GAPDH.Mean ± SEM, n = 3. F) Immunohistochemical staining for pSTAT1 in PC3 METTL1 KO and WT cell-derived xenografts.Scale bar represents 50 μm.G, H) Increased mRNA expression levels of ISGs genes in METTL1 KO PC3 (G) and METTL1-silenced 22Rv1 cells (H) cells.Mean ± SD, n = 3. I) Western blot of IRF9 and METTL1 protein levels in METTL1-silenced 22Rv1 cells compared to control cells (SCR).Bottom bar plots show quantification of IRF9 expression normalised to GAPDH.Mean ± SD, n = 3. J) Correlation analysis between METTL1 expression and ISG genes in three human PCa expression datasets reflects an inverse expression correlation.Correlations with p-value < = 0.05 and |R|> = 0.2 are indicated with (*).Statistical tests: One-tailed (B, G-I) and two-tailed Student's t-test and Pearson correlation test (J) were performed.*p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.Additional file 8: Supplementary Figure S8.Immune cell characterisation and correlation with METTL1 expression in human PCa.A) Complete cytokine array analysis of PC3 METTL1 KO compared to WT cells' c.m. shows differential expression of inflammatory cytokines.Mean ± SD of log 2 fold change, n = 4. B) tSNE analysis of the expression markers of M1-, M2-like and Mø macrophages exposed to PC3 METTL1 KO and WT cells' c.m. of a second biological replicate (KO2 and WT2).C) Phagocytic capacity of macrophages exposed to PC3 METTL1 KO and WT cells' c.m. Mean ± SEM, n = 3. D) Polarisation of primary peripheral blood monocytes (PMBCs) exposed to PC3 METTL1 KO and WT cells' c.m.The upper panel shows a schematic of the workflow used.The lower panels show the quantification of the percentage of positive cells.n = 3 technical replicates of 2 biological replicates.E) Activation of T cells from a second blood donor and coculture with macrophages exposed to PC3 WT and METTL1 KO cells' c.m. T cells were stained with CFSE on day 0 and measured on day 3. Means ± SD, n = 3 for two biological replicates.F) Representative immunohistochemical staining for METTL1 and CD86 (M1-like macrophages), CD163 (M2-likemacrophages), and CD8 T cells in human PCa biopsies.The scale bar represents 50 μm.G, H) METTL1 mRNA expression inversely correlates with M1-like macrophage infiltration using the CIBERSORT deconvolution algorithm with TCGA and Taylor datasets (G), and TAM infiltration analysed using TIMER (H).Statistical tests: One-tailed t-test: *p < 0.05, **p < 0.01, ***p < 0.001 (D, E), Pearson's correlation (r), p-value (pV), and linear regression with 95% confidence (bands) are shown (G, H).Additional file 9: Supplementary Figure S9.Characterization of Mettl1 deletion in murine PCa.A) Schematic overview of Mettl1 systemic knockout and Mettl1 flox/flox mouse model generation.B) mRNA levels of Mettl1 in prostatic tissue from five-month-old Pten-KO/Mettl1+/+, Pten-KO/ Mettl1+/-, and Pten-KO/Mettl1 fl/fl mice.Mean ± SD, n=3.C) 5'tRNA fragment accumulation in Pten-KO/Mettl1+/+, Pten-KO/Mettl1+/-and Pten-KO/ Mettl1 fl/fl tumours.The right bar chart shows means ± SD, n ≥5.D) Haematoxylin and eosin staining of Pten-KO/Mettl1+/+, Pten-KO/Mettl1+/-, and Pten-KO/Mettl1 fl/fl tumour sections from the ventral, dorsal, and anterior lobes.E) Representative images of immunostained sections for METTL1 and markers of proliferation (Ki67), M1-like (iNOS) and M2-like (CD68, Arg1) macrophages, and CD8+ T cells (CD8) from Pten-KO/Mettl1+/+, Pten-KO/ Mettl1+/-, and Pten-KO/Mettl1 fl/fl prostate tumours.Arrows indicate positive cells.F) Prostate tumour volume from Pten-KO/Mettl1+/+ and Pten-KO/Mettl1+/-five-month-old mice reflects reduced tumour burden after Mettl1 deletion.Ventral (V), dorsal (D) and anterior (A) lobes.Mean ± SD, n≥3.G) Systemic deletion of Mettl1 resulted in reduced tumour proliferation (Ki67+ cells) and increased immune infiltration of iNOS+ (M1-like) macrophages and CD8+ T cells.Staining of tumours from Pten-KO/Mettl1+/+ and Pten-KO/Mettl1+/-five-month-old mice.Mean ± SD, n≥3, >10 images per biological replicate.H) Tumour growth of allografts derived from Pten-KO/Mettl1+/+ and Pten-KO/Mettl1 flox/flox tumour cells , unexposed (0 h) or after 2 and 8 h of oxidative stress exposure.The loading control by red-safe staining (tRNA) is shown in the bottom panels (B, C, F-H, J).Bands corresponding to full-length tRNAs are indicated with stars, and 5'tRNA fragments are indicated with arrows.Statistical tests: One-tailed Student's t-test (C-F).*p < 0.05.