The FAM86 domain of FAM86A confers substrate specificity to promote EEF2-Lys525 methylation

• DOI: 10.1016/j.jbc.2023.104842
• Publication Date: 2023-05-18T15:12:37Z
• Authors: Joel William Francis; Zengyu Shao; Pradnya Narkhede; Annie Truc Trinh; Jiuwei Lu; Jikui Song; Or Gozani
• Keywords: Models, Molecular; 570; Protein Structure; S-Adenosylmethionine; mRNA translation; protein synthesis; 610; Crystallography, X-Ray; Methylation; Substrate Specificity; Peptide Elongation Factor 2; Protein Domains; Models; FAM86A; Humans; Point Mutation; Crystallography; Lysine; EEF2; Molecular; Methyltransferases; EEF2KMT; 3. Good health; alphafold; translation elongation; Protein Structure, Tertiary; lysine methylation; ribosome; X-Ray; methyltransferase; Tertiary; Research Article

FAM86A is a class I lysine methyltransferase (KMT) that generates trimethylation on the eukaryotic translation elongation factor 2 (EEF2) at Lys525. Publicly available data from The Cancer Dependency Map project indicate high dependence of hundreds human cancer cell lines expression. This classifies among numerous other KMTs as potential targets for future anticancer therapies. However, selective inhibition by small molecules can be challenging due to conservation within S-adenosyl methionine (SAM) cofactor binding domain KMT subfamilies. Therefore, understanding unique interactions each KMT–substrate pair facilitate developing highly specific inhibitors. gene encodes an N-terminal FAM86 unknown function in addition its C-terminal domain. Here, we used combination X-ray crystallography, AlphaFold algorithms, and experimental biochemistry identify essential role mediating EEF2 methylation FAM86A. To our studies, also generated EEF2K525 methyl antibody. Overall, this first report biological structural any species example noncatalytic participating protein methylation. interaction between provides new strategy molecule inhibitor, results provide which modeling protein–protein with expedites biology. Protein one, two, or three moieties ε-nitrogen side chain, forming mono-, di-, trimethylated derivatives (1Bhat K.P. Umit Kaniskan H. Jin J. Gozani O. Epigenetics beyond: targeting writers treat disease.Nat. Rev. Drug Discov. 2021; 20: 265-286Crossref PubMed Scopus (77) Google Scholar). Lysine abundant posttranslational modification humans, carried out dozens methyltransferases (KMTs) encoded genome Scholar, 2Carlson S.M. Nonhistone regulation pathways.Cold Spring Harb. Perspect. Med. 2016; 6https://doi.org/10.1101/cshperspect.a026435Crossref (78) 3Falnes P. Jakobsson M.E. Davydova E. Ho A. Małecki seven-β-strand methyltransferases.Biochem. 473: 1995-2009Crossref (63) Human belong one two large families: SET (Su(var)3–9, Enhancer-of-zeste, Trithorax) enzymes 7βS (seven-β-strand) (4Clarke S.G. surface buried deep: thinking outside histone box trends.Biochem. Sci. 2013; 38: 243-252Abstract Full Text PDF (134) Members family are known generate histones well nonhistone proteins serve both nuclear cytoplasmic functions (5Husmann D. Histone biology Struct. Mol. Biol. 2019; 26: 880-889Crossref (168) family, few exceptions (6Vlaming van Leeuwen F. upstreams downstreams H3K79 methylation.DOT1L Chromosoma. 125: 593-605Crossref (64) 7Metzger Wang S. Urban Willmann Schmidt Offermann et al.KMT9 monomethylates H4 12 controls proliferation prostate cells.Nat. 361-371Crossref (46) Scholar), generally methylate activities (3Falnes regulate diverse processes via their activities, including fundamental transcription 4Clarke 5Husmann Accordingly, dysregulation many KMTs, aberrant expression mutation, has been linked diseases member 525 (K525me3) (8Davydova A.Y.Y. Malecki Moen Enserink J.M. al.Identification characterization novel evolutionarily conserved lysine-specific (eEF2).J. Chem. 2014; 289: 30499-30510Abstract (36) 9Zhang L. Hamey J.J. Hart-Smith G. Erce M.A. Wilkins M.R. Elongation 3--a methyltransferase.Biochem. Biophys. Res. Commun. 451: 229-234Crossref (15) performs facilitating ribosomal translocation during mRNA (10Schuller A.P. Green R. Roadblocks resolutions translation.Nat. Cell 2018; 19: 526-541Crossref (113) 11Lucas-Lenard Lipmann Separation microbial amino acid polymerization factors.Proc. Natl. Acad. U. 1966; 55: 1562-1566Crossref 12Taylor D.J. Nilsson Merrill A.R. Andersen G.R. Nissen Frank Structures modified eEF2 80S ribosome complexes reveal GTP hydrolysis translocation.EMBO 2007; 2421-2431Crossref (150) 13Kaul Pattan Rafeequi T. Eukaryotic factor-2 (eEF2): peptide chain cell.Biochem. Funct. 2011; 29: 227-234Crossref (142) It reported translational control proteome, upregulation (14Nakamura Aoyagi Nanchi I. Nakatsuka Hirata Shibata al.Overexpression gastrointestinal cancers involvement G2/M progression cycle.Int. Oncol. 2009; 34: 1181-1189PubMed 15Oji Y. Tatsumi N. Fukuda M. al.The tumor-associated antigen overexpressed various types cancers.Int. 44: 1461-1469Crossref (52) common (16Bhat Robichaud Hulea Sonenberg Pelletier Topisirovic Targeting machinery cancer.Nat. 2015; 14: 261-278Crossref (522) 17Silvera Formenti S.C. Schneider R.J. Translational Cancer. 2010; 10: 254-266Crossref (630) Notably, publicly (DepMap) (Fig. S1) (18Meyers R.M. Bryan J.G. McFarland Weir B.A. Sizemore A.E. Xu al.Computational correction copy number effect improves specificity CRISPR-Cas9 essentiality screens Genet. 2017; 49: 1779-1784Crossref (880) 19Dempster Rossen Kazachkova Pan Kugener Root D.E. al.extracting insights achilles genome-scale CRISPR lines.bioRxiv. ([preprint])https://doi.org/10.1101/720243Crossref (0) observation suggests there may value oncological indications. While success last decade inhibitors, advancing inhibitors into successful clinical trials ongoing challenge 20Luo Chemical biochemical perspectives methylation.Chem. 118: 6656-6705Crossref (114) 21Schapira Arrowsmith C.H. Methyltransferase modulation epigenome beyond.Curr. Opin. 33: 81-87Crossref (19) Many under investigation use competitors S-adenosylmethionine (SAM), donor catalysis inhibiting SAM-competitive effective, these carry poor selectivity since all utilize SAM catalysis. Thus, identifying characteristics pairs aid development encode domains catalytic (e.g., (22Ren W. Lu Huang Gao Li G.G. al.Structure ZCCHC4 m(6)A-methylation 28S rRNA.Nat. 5042Crossref (55) Scholar)). In Saccharomyces cerevisiae uncharacterized (MTase) 1A). Primary sequence alignment shows MTase yeast 20% 28% identity, respectively Although evolved alongside through evolutionary tree, remains no species. machine learning algorithm developed DeepMind predict structures based primary sequences (23Jumper Evans Pritzel Figurnov Ronneberger al.Highly accurate structure prediction AlphaFold.Nature. 596: 583-589Crossref (7365) couple years, thousands models have predicted made public (24Cramer AlphaFold2 biology.Nat. 28: 704-705Crossref (98) These often comparisons experimentally determined 25Stevens A.O. He Benchmarking accuracy loop prediction.Biomolecules. 2022; 12https://doi.org/10.3390/biom12070985Crossref (6) AlphaFold-Multimer iteration designed possible conformations multichain (26Evans O'Neill Antropova Senior al.Protein complex AlphaFold-Multimer.bioRxiv. ([preprint])https://doi.org/10.1101/2021.10.04.463034Crossref simulation silico process enabling molecular accelerated timescale. work, newest characterize Using provided crystallography AlphaFold, forms five-helix bundle distinct By testing activity truncated construct lacking domain, found required vitro cells. Simulated FAM86A–EEF2 suggested extensive intermolecular EEF2. Point mutations interface inhibited Based data, propose confers substrate orienting EEF2-Lys525 toward active site. Future work leverage take similar approaches develop KMTs. Our initial effort crystallizing wildtype failed diffractable crystals. overcome challenge, harboring triple mutation reduction entropy (see Experimental procedures), permitted determination full-length cofactor–by-product homocysteine (SAH) 3.3-Å resolution (Figs. 1, B C, S2 Table 1). crystal FAM86A–SAH belongs I422 space group, containing asymmetric unit S2A). We were able trace entire protein, except very N terminus (residues 1–5) portion linker 127–133).Table 1Crystallographic collection refinement statisticsFAM86–SAH (PDB 8FZB)Data Space groupI 4 dimensionsa, b, c (Å)160.7, 160.7, 351.8α, β, γ (°)90, 90, 90 Wavelength0.9792 Resolution (Å)48.72–3.33 (3.45–3.33)aValues parentheses highest-resolution shell. dataset was collected single crystal. Rmerge0.149 (1.64) I/σI19.8 (1.8) CC1/20.999 (0.359) Completeness (%)98.6 (96.6) Redundancy39 (34) Total reflections1,324,854 (112,423) Unique reflections33,996 (3229)Refinement No. reflections33,573 Rwork/Rfree (%)23.0/25.3 atomsProtein7300Ligands78 factors (Å2)Protein139.8Ligands104.5 RMS deviationsBond lengths (Å)0.003Bond angles (°)0.67 Ramachandran Favored (%)96.85 Allowed (%)3.15 Outliers (%)0a Values Open table tab reveals two-lobe architecture, stacks right top creating cleft commonly observed DNA C) (27Cheng X. Structure methyltransferases.Annu. Biomol. 1995; 24: 293-318Crossref 28Ren Song Structural basis DNMT1 DNMT3A-mediated methylation.Genes (Basel). 9https://doi.org/10.3390/genes9120620Crossref (62) dominated (αa, αb, αc, αe, αf), packs against following two-stranded (β1 β2) β-sheet αa- intervening αd-helix. assumes Rossmann fold, seven-stranded central flanked α-helices sides. next examined model FAM86A's 3D database ligand transposition AlphaFill (29Hekkelman M.L. de Vries Joosten R.P. Perrakis AlphaFill: enriching ligands cofactors.Nat. Met. 2023; 205-213Crossref (21) bears striking resemblance structure, root-mean-square deviation (RMSD) 1.2 Å over 319 aligned Cα atoms 1D). Of note, nearly identical SAM-binding pocket approximately 14 residues coordination SAH 1D) Nevertheless, show slight difference helical orientations (αe αf) positionings SAH-engaging W139 Y251 did not interact began raising antibody EEF2-K525 S3). First, knockout (KO) HEK293T cells endogenous EEF2-K525me3 levels completely depleted KO 2A). custom ask whether test this, cloned (FAM86A(101–330)) 2B) tested adding back transfection. EEF2-K525me3, mutant, expressed same level 2C). Next, asked question incubating recombinant, purified presence radiolabeled (3H-SAM). Consistent result cells, 2D). When FAM86A(101–330) incubated whole extracts vitro, observe whereas methylated size 2E). From conclude likely main physiologic target considered participates physical silico, queried ColabFold (30Mirdita Schütze K. Moriwaki Heo Ovchinnikov Steinegger ColabFold: making folding accessible all.Nat. 679-682Crossref (811) Scholar) simulate heterodimeric 1:1 stoichiometry. outputs multiple other, assigning confidence ranking model. query another, designated highest analyzed here. A less than 3 predicted, involving 3A). Importantly, occupy position proximal site near independently identified superimposed model, visualize proximity if FAM86A, EEF2, cocrystallized 3B). second interface, IV close-up view network Ile78, Pro88, Asp90, Try93, Glu94, Leu96, Ala97, Leu100, Met101 Pro596, Val680, Ala681, Trp685, Gly718, Gly719, Gln720, Ile722, Pro723, Arg726 3C). Since folds goal simulating interaction, individually onto measure quality control. individual conformation complex, RMSD only 0.366 252 3D). case calculated 0.733 642 3E). Most variation caused rotation IV, flexibility crucial (12Taylor 31Spahn C.M. Gomez-Lorenzo M.G. Grassucci R.A. Jørgensen Beckmann al.Domain movements tRNA 2004; 23: 1008-1019Crossref (332) Given degree similarity protein's viewed reasonable further experiments. hypothesized accurately portrays orients blocking inhibits FAM86A-mediated model-guided carrying point respectively. Specifically, FAM86A-Ala97 EEF2-Ile722 substituted arginine increase likelihood disrupting while still maintaining native S4, A–D). As shown Figure 4A, complementation (A97R) rescue restored 4A). EEF2(I722R) viable transfected immunoprecipitated FLAG-EEF2(I722R) endogenously expressing input sample I722R sample, exogenous substitution 4B). mutant constructs equally compared impact overall 4, C D). suggest substitutions prevent allosteric mechanism, AlphaFold-Multimer. 4C). contrast, showed reduced 4D). postulate that, physiological context, transient nature becomes sensitive modest interference mutations. Meanwhile, simplicity stoichiometry reaction mixtures amplify possibility might another. implication intriguing fronts. instance Second, represents clinically actionable activity, relevant given axis Third, domain–containing genes emerged ancestor. exists yeast, neither FAM86B1 FAM86B2 lower animals mice (32Bult C.J. Blake J.A. Smith C.L. Kadin andRichardson J.E. Mouse (MGD) 2019.Nucl. Acids 47: D801-d806Crossref (399) members substrates. Alignment FAM86B1, homology genes, implying close relationship 5, B). resulting loss exon (33Rosenbloom K.R. Sloan C.A. Malladi V.S. Dreszer T.R. Learned Kirkup V.M. al.ENCODE UCSC browser: year 5 update.Nucl. 41: D56-63Crossref (584) form bundles, four-helix 5C). lost homologous helix redundant copies nor 5D), suggesting they different substrates enzymes. previously humans. Here process. extremely FAM86B2, suspect but cannot inactive duplicate DepMap support hypothesis revealing expression, contrast S1 S5A). categorized S5B). it could enzyme hints interesting story: deleting promote methylation, like axis, promotes critical (34Yu Chojnowski Rosenthal Kosinski AlphaPulldown-a python package protein-protein using AlphaFold-Multimer.Bioinformatics. 39https://doi.org/10.1093/bioinformatics/btac749Crossref (2) tool FAM86B1. study describes artificial intelligence direct truncation experiments determine clear testable hy