METTL9 mediated N1-histidine methylation of zinc transporters is required for tumor growth

Histidine methylation has been known for many years (Searle and Westall, 1951), but only a few proteins carrying such modifications have been studied (Webb et al., 2010; AlHadid et al., 2014; Kwiatkowski et al., 2018; Guo et al., 2019; Wilkinson et al., 2019; Kwiatkowski and Drozak, 2020). Recent proteomic studies suggest that more than 13% of all protein methylation events in human methylome was attributed to the modification of protein histidine residues (Ning et al., 2016). Histidine can be methylated at either N1 or N3 position of its imidazole ring, yielding the isomers 1-methylhistidine (His(1-me)) or 3-methylhistidine (His(3-me)). So far, only a limited number of histidine methyltransferases have been reported. Hpm1 and SETD3 modify N3-methylhisitidine on specific substrates, Rpl3 and actin, respectively (Webb et al., 2010; Wilkinson et al., 2019). CARNMT1 is an N1 position-specific methyltransferase that catalyzes dipeptide (Cao et al., 2018). However, the methyltransferases that modify N1-methylhisitidine on protein substrates, as well as the functional and (patho)physiological significance of such modification, still remain a major knowledge gap in the field of protein posttranslational modification. Recent findings highlight methyltransferase-like protein family (METTL) as an important family of putative methyltransferase (Jiang et al., 2021). As Mettl9 is broadly expressed in various cancer cell lines and its biological functions are unknown, we set out to study Mettl9. The targeting strategy of knocked-out Mettl9 and the gene knockout efficiency were evaluated (Fig. S1A–C). RM-1 cells lack of Mettl9 (hereafter termed Mettl9 KO cells) showed significantly reduced growth and colony formation in vitro (Fig. 1A– C). Similar results were observed in Mettl9 KO MC38 cells (Fig. S1D). Next, we found that deletion of Mettl9 significantly inhibited tumor autonomous growth in vivo (Fig. 1D– F). Consistent with what we observed in immunodeficient mice, Mettl9 ablation similarly resulted in compromised tumor growth in immunocompetent mice (Fig. 1G–I). Analysis of intratumoral immune cells showed that both the numbers and percentages of CD45 cells were significantly increased in Mettl9 KO tumors compared to WT tumors (Fig. S1E and S1F). Among intratumoral immune cells, the percentages of CD4 and CD8 effector T cells were significantly increased (Fig. S1G and S1H). Collectively, these results indicated that Mettl9 deletion not only suppresses tumor cell autonomous growth, but also elicits potent antitumor immunity to restrain tumor burden. As we know, METTL9 belongs to the seven-β-strand (7BS) methyltransferase, characterized by a twisted beta-sheet structure and certain conserved sequence motifs (Petrossian and Clarke, 2011). We aligned the sequence of METTL9 in different species, and found that METTL9 has conserved sequence motifs that are similar to METTL family members that are responsible for proteinmethylation (Fig. S2A).We first demonstrated that purified recombinant mouse METTL9 protein could not methylate a histone substrate (Fig. S3A). Considering the fact that METTL9 protein was localized in the cytoplasm (Fig. S3B), we detected the activity of METTL9 to methylate proteins inWTandMettl9KO tumor cell lysates and observed a substrate protein approximately 55-kDa was methylated (Fig. 1J), indicating that METTL9 indeed has enzymatic activity and is a non-histone protein methyltransferase. We next constructedMettl9-3×flag knock-in RM-1 cell line and identified several interactants of endogenous METTL9 in tumor cells (Fig. S3C). Among them,SLC39A7 is a previously published METTL9 interactant (Ignatova et al., 2019), which is a zinc transporter, nearly 55KD and abundantly expressed in cell lysates (Fig. S3D).We next confirmed SLC39A7 was indeed an in vitro substrate of METTL9 (Fig. 1K). To fine-mapping of the methylation sites at SLC39A7, we performed in vitro METTL9 methylation assay on multiple truncated peptides (Figs. 1L and 1M, S3E and S3F). Two of them that could be methylated by METTL9 were further identified bymass spectrum to pinpoint the methylated residues. The results showed only histidine residues in these peptides were methylated (Fig. S3G and S3H). We observed that replacement of histidine by alanine abolished METTL9 mediated methylation (Fig. 1N). Correspondingly, although METTL9 could not methylate histone itself, it could indeed methylate histoneH3with a 6×His tag (Fig. S4A). Aswe know, histidine can potentially be methylated on the nitrogen in position 1 (π, N1) or 3 (τ, N3) of the imidazole ring, and


Supplemental Methods
Cell culture.
The clones were screened by genomic PCR. The flag KI clones were screened by genomic PCR and confirmed by Western Blot using anti-flag antibody.
The membrane was explored by using enhanced chemiluminescence kit (Biosharp) according to the manufacturer's instructions.

Real-time RT-PCR analysis.
TRNzol (TIANGEN) was used to extract RNA from cells growing on tissue culture dishes. 500 ng RNA were used for following RT-PCR. The HiScript RT SuperMix kit CCACTTGCCACCTACGTTTT.

Cell proliferation and colony formation assay.
Cell proliferation was quantified using the cell counting Kit-8 (CCK8; APExBIO) according to the manufacturer's instruction. Cells were seeded in 96-well plates in triplicate at an initial density of 2000 cells per well. After being cultured for 24, 48, 72, and 96 hours, CCK-8 solution in 10 ul was added to each well, and cells were further incubated at 37 °C for 4 hours. Optical density (OD) value was measured spectrophotometrically at 450 nm wavelength.
Cells were plated in 6-well plates (10000 cells per well), and then cultured for 4 days.
The overexpression cells were generated by infecting with BFP-Slc39a7 or mutants' virus.
After 24 h, the BFP-positive cells were sorted by flow cytometry. For an assay of colony formation, colonies were fixed with 4% paraformaldehyde for 20 minutes, stained with 0.1% crystal violet for 15 minutes, and then washed with phosphate-buffered saline (PBS) before measurement. The total colony area was measured by image J software. The relative area was calculated by total area divided by original cell number.
Tumor formation assay.
The C57B6/J mice and BALB/c nude mice were feeding under specific pathogen-free conditions in the experimental animal department of USTC. For the in vivo tumor formation assays, 1.5×10 5 RM-1 Mettl9 KO cells (1#, 2#) or RM-1 wild type cells were subcutaneously injected into the C57B6/J mice (n = 5 for each group) and the BALB/c nude mouse (n = 6 for each group) at 8 weeks of age. After transplantation 8 days, the growth of the tumors was assessed every two days. The mice were sacrificed after a period of about 3 weeks, and the weights of subcutaneous tumors were measured.

Analysis of immune cell infiltrate in tumor microenvironment.
Preparation of single-cell suspension of mouse tumor tissues by mechanical grinding and collagenase digestion. For flow cytometry analysis, after blocking, cells were stained with the following mAbs: anti-CD45, anti-CD4, anti-CD8a, all from Biolegend. Samples were run on a BD flow cytometer and analyzed with the Flowjo Analysis Software (Beckman Coulter, Pasadena, CA). Cell populations were defined as follows: CD45 + CD4 + CD8 -(CD4 + T cells); CD45 + CD4 -CD8 + (CD8 + T cells).

IP assay and Mass spectrometry.
The 15 cm dish Mettl9-3×flag RM-1 cells extracts prepared with IP buffer (25 mM Tris-HCl pH 7.4, 150 mM NaCl, 1 mM EDTA,1% NP-40, 5% glycerol) remain on the ice for 30 minutes. Lysates were cleared by centrifugation at 12000 rpm for 10 minutes and incubated with anti-flag (anti-flag M2 Affinity Gel, Sigma) antibodies agarose. after 4 hours beads were then collected, and then the mass spectrum was performed. To select the METTL9-interacting proteins, the PEMs and Sum PEP Score was utilized to do a dot plot and top proteins closest to the upper right corner (PSMs > 2; log2(Sum PEP Score) > 2) was selected. For synthetic peptides methylation detection, after methylated by METTL9, the sample were sent to MS detection.
In vitro methyltransferase assays.
In vitro methyltransferase assays were performed using 30 ul reaction buffer ( analysis, the peptides were directly loaded to NC membranes and exposed to XBT X-ray film (Carestream) for 7 days.

Fluorescence imaging of Zn 2+ level.
Cell were cultured in the glass bottom micro-well dishes (Biosharp) for confocal microscopy imaging. Before ratiometric imaging with LSM 710 microscopy, 2 uM FluoZin-3 AM (Invitrogen) was add to dishes to stain cells for 1 hour and then was washed twice with PBS. For overexpression, the cells in 10 cm dish were infected with Slc39a7 wildtype or mutants in Plvx-IRES-BFP virus. After 24 h, the BFP-positive cells were sorted by flow cytometry. After cultured in the glass bottom micro-well dishes for 24 hours, the cells were stained using FluoZin-3 AM (2 uM) for confocal microscopy imaging.

RNA-seq data processing and analysis.
The RM-1 cells of 1×10 7 WT and Mettl9 KO were extracted with TRNzol (TIANGEN) for total RNA library preparation. The libraries are directly sequenced using next-generation sequencing technologies. After filtering, clean reads were mapped to the mouse reference genome. Mapping results were stored in BAM files. Total read counts at the gene level were summarized using feature-counts function in R environment, with the R package biomaRt for gene and transcript mapping. The differential expression genes were analyzed by DESeq2 with default settings using total read counts as input and the adjusted P value (p.adj) less than 0.05. Dotplot of differential genes of normalized expression matrix from DESeq2 analysis from GO pathway enrichment.
We performed gene set enrichment analysis (GSEA) using KEGG function of clusterProfiler package in RStudio. The RNA-seq data of non-targeting control (NC) and two individual Slc39a7 siRNA in MDA-MB-231 cells are from the accession number GSE155437 from NCBI Gene Expression Omnibus (PMID: 33608508). Significant KEGG pathways with an enrichment score > 0.4 and a p-value <0.01 were identified. The R package ggplot2 was applied to visualize the results.

TCGA and GTEx data analysis.
We download the expression data of METTL9 in PRAD, PAAD, LIHC combing GTEx normal sample from USCS Xena (http://xena.ucsc.edu), and make boxplot with significance difference index through ggplot in R environment.
We applied ESTIMATE computational method to calculate the immune score of different cancer samples from The Cancer Genome Atlas (TCGA) database. Correlation between several genes expression levels and immune scores were performed using the 'rcorr' function implemented in Hmisc R package v.4.5.0, adopting Pearson's correlation method.
Significant correlations (P<0.05) were plotted using the pheatmap R package v.1.0.12. Genes with no notable correlation show a coefficient equal to zero.

Statistical analysis.
All experimental operations were independently repeated for at least three times.
Images was analyzed using Image J Software (National Institutes of Health, Bethesda, MD, USA). Statistical analysis was performed using GraphPad Prism Software 6 (GraphPad Software, La Jolla, CA, USA). For data that had a normal distribution and homogeneity of variance, two-tailed Student's t test was performed to evaluate significant differences between two groups. P-values<0.05 were considered statistically significant.