Turning Nonselective Inhibitors of Type I Protein Arginine Methyltransferases into Potent and Selective Inhibitors of Protein Arginine Methyltransferase 4 through a Deconstruction-Reconstruction and Fragment-Growing Approach.

Protein arginine methyltransferases (PRMTs) are important therapeutic targets, playing a crucial role in the regulation of many cellular processes and being linked to many diseases. Yet, there is still much to be understood regarding their functions and the biological pathways in which they are involved, as well as on the structural requirements that could drive the development of selective modulators of PRMT activity. Here we report a deconstruction-reconstruction approach that, starting from a series of type I PRMT inhibitors previously identified by us, allowed for the identification of potent and selective inhibitors of PRMT4, which regardless of the low cell permeability show an evident reduction of arginine methylation levels in MCF7 cells and a marked reduction of proliferation. We also report crystal structures with various PRMTs supporting the observed specificity and selectivity.

Hijacking DNA methyltransferase transition state analogues to produce chemical scaffolds for PRMT inhibitors.

DNA, RNA and histone methylation is implicated in various human diseases such as cancer or viral infections, playing a major role in cell process regulation, especially in modulation of gene expression. Here we developed a convergent synthetic pathway starting from a protected bromomethylcytosine derivative to synthesize transition state analogues of the DNA methyltransferases. This approach led to seven 5-methylcytosine-adenosine compounds that were, surprisingly, inactive against hDNMT1, hDNMT3Acat, TRDMT1 and other RNA human and viral methyltransferases. Interestingly, compound 4 and its derivative 2 showed an inhibitory activity against PRMT4 in the micromolar range. Crystal structures showed that compound 4 binds to the PRMT4 active site, displacing strongly the S-adenosyl-l-methionine cofactor, occupying its binding site, and interacting with the arginine substrate site through the cytosine moiety that probes the space filled by a substrate peptide methylation intermediate. Furthermore, the binding of the compounds induces important structural switches. These findings open new routes for the conception of new potent PRMT4 inhibitors based on the 5-methylcytosine-adenosine scaffold.This article is part of a discussion meeting issue 'Frontiers in epigenetic chemical biology'.

Structural studies of protein arginine methyltransferase 2 reveal its interactions with potential substrates and inhibitors.

PRMT2 is the less-characterized member of the protein arginine methyltransferase family in terms of structure, activity, and cellular functions. PRMT2 is a modular protein containing a catalytic Ado-Met-binding domain and unique Src homology 3 domain that binds proteins with proline-rich motifs. PRMT2 is involved in a variety of cellular processes and has diverse roles in transcriptional regulation through different mechanisms depending on its binding partners. PRMT2 has been demonstrated to have weak methyltransferase activity on a histone H4 substrate, but its optimal substrates have not yet been identified. To obtain insights into the function and activity of PRMT2, we solve several crystal structures of PRMT2 from two homologs (zebrafish and mouse) in complex with either the methylation product S-adenosyl-L-homocysteine or other compounds including the first synthetic PRMT2 inhibitor (Cp1) studied so far. We reveal that the N-terminal-containing SH3 module is disordered in the full-length crystal structures, and highlights idiosyncratic features of the PRMT2 active site. We identify a new nonhistone protein substrate belonging to the serine-/arginine-rich protein family which interacts with PRMT2 and we characterize six methylation sites by mass spectrometry. To better understand structural basis for Cp1 binding, we also solve the structure of the complex PRMT4:Cp1. We compare the inhibitor-protein interactions occurring in the PRMT2 and PRMT4 complex crystal structures and show that this compound inhibits efficiently PRMT2. These results are a first step toward a better understanding of PRMT2 substrate recognition and may accelerate the development of structure-based drug design of PRMT2 inhibitors. DATABASE: All coordinates and structure factors have been deposited in the Protein Data Bank: zPRMT21-408 -SFG = 5g02; zPRMT273-408 -SAH = 5fub; mPRMT21-445 -SAH = 5ful; mPRMT21-445 -Cp1 = 5fwa, mCARM1130-487 -Cp1 = 5k8v.

Functional insights from high resolution structures of mouse protein arginine methyltransferase 6.

PRMT6 is a protein arginine methyltransferase involved in transcriptional regulation, human immunodeficiency virus pathogenesis, DNA base excision repair, and cell cycle progression. Like other PRMTs, PRMT6 is overexpressed in several cancer types and is therefore considered as a potential anti-cancer drug target. In the present study, we described six crystal structures of PRMT6 from Mus musculus, solved and refined at 1.34 Å for the highest resolution structure. The crystal structures revealed that the folding of the helix αX is required to stabilize a productive active site before methylation of the bound peptide can occur. In the absence of cofactor, metal cations can be found in the catalytic pocket at the expected position of the guanidinium moiety of the target arginine substrate. Using mass spectrometry under native conditions, we show that PRMT6 dimer binds two cofactor and a single H4 peptide molecules. Finally, we characterized a new site of in vitro automethylation of mouse PRMT6 at position 7.

Crystal structures of Entamoeba histolytica lysyl-tRNA synthetase reveal conformational changes upon lysine binding and a specific helix bundle domain.

The class II lysyl-tRNA synthetases (KRS) are conserved aminoacyl-tRNA synthetases that attach lysine to the cognate tRNA in a two-step mechanism. The enzyme from the parasitic protozoan Entamoeba histolytica was crystallized in the presence of small ligands to generate snapshots of the lysine-adenylate formation. The residues involved in lysine activation are highly conserved and the active site closes around the lysyl-adenylate, as observed in bacterial KRS. The Entamoeba EMAPII-like polypeptide is not resolved in the crystals, but another Entamoeba-specific insertion could be modeled as a small helix bundle that may contribute to tRNA binding through interaction with the tRNA hinge.

Structural insight into arginine methylation by the mouse protein arginine methyltransferase 7: a zinc finger freezes the mimic of the dimeric state into a single active site.

Protein arginine methyltransferase 7 (PRMT7) is a type III arginine methyltransferase which has been implicated in several biological processes such as transcriptional regulation, DNA damage repair, RNA splicing, cell differentiation and metastasis. PRMT7 is a unique but less characterized member of the family of PRMTs. The crystal structure of full-length PRMT7 from Mus musculus refined at 1.7 Šresolution is described. The PRMT7 structure is composed of two catalytic modules in tandem forming a pseudo-dimer and contains only one AdoHcy molecule bound to the N-terminal module. The high-resolution crystal structure presented here revealed several structural features showing that the second active site is frozen in an inactive state by a conserved zinc finger located at the junction between the two PRMT modules and by the collapse of two degenerated AdoMet-binding loops.

Cloning, expression, purification and preliminary X-ray crystallographic analysis of mouse protein arginine methyltransferase 7.

Protein arginine methyltransferase 7 (PRMT7) is a unique but less characterized member of the family of protein arginine methyltransferases (PRMTs) that plays a role in male germline gene imprinting. PRMT7 is the only known PRMT member that catalyzes the monomethylation but not the dimethylation of the target arginine residues and harbours two catalytic domains in tandem. PRMT7 genes from five different species were cloned and expressed in Escherichia coli and Sf21 insect cells. Four gave soluble proteins from Sf21 cells, of which two were homogeneous and one gave crystals. The mouse PRMT7 structure was solved by the single anomalous dispersion method using a crystal soaked with thimerosal that diffracted to beyond 2.1 Šresolution. The crystal belonged to space group P4(3)2(1)2, with unit-cell parameters a = b = 97.4, c = 168.1 Šand one PRMT7 monomer in the asymmetric unit. The structure of another crystal form belonging to space group I222 was solved by molecular replacement.

Expression, purification, crystallization and preliminary X-ray crystallographic analysis of Zucchini from Drosophila melanogaster.

PIWI-interacting RNAs (piRNAs) bind PIWI proteins and silence transposons to maintain the genomic integrity of germ cells. Zucchini (Zuc), a phospholipase D superfamily member, is conserved among animals and is implicated in piRNA biogenesis. However, the underlying mechanism by which Zuc participates in piRNA biogenesis remains elusive. Drosophila melanogaster Zuc (DmZuc) was expressed in Escherichia coli, purified and crystallized. X-ray diffraction data were collected to 1.75 Šresolution. The crystal belonged to space group P2(1), with unit-cell parameters a=55.0, b=71.2, c=56.3 Å, ß=107.9°.

Structure and function of Zucchini endoribonuclease in piRNA biogenesis.

PIWI-interacting RNAs (piRNAs) silence transposons to maintain genome integrity in animal germ lines. piRNAs are classified as primary and secondary piRNAs, depending on their biogenesis machinery. Primary piRNAs are processed from long non-coding RNA precursors transcribed from piRNA clusters in the genome through the primary processing pathway. Although the existence of a ribonuclease participating in this pathway has been predicted, its molecular identity remained unknown. Here we show that Zucchini (Zuc), a mitochondrial phospholipase D (PLD) superfamily member, is an endoribonuclease essential for primary piRNA biogenesis. We solved the crystal structure of Drosophila melanogaster Zuc (DmZuc) at 1.75 Å resolution. The structure revealed that DmZuc has a positively charged, narrow catalytic groove at the dimer interface, which could accommodate a single-stranded, but not a double-stranded, RNA. DmZuc and the mouse homologue MmZuc (also known as Pld6 and MitoPLD) showed endoribonuclease activity for single-stranded RNAs in vitro. The RNA cleavage products bear a 5'-monophosphate group, a hallmark of mature piRNAs. Mutational analyses revealed that the conserved active-site residues of DmZuc are critical for the ribonuclease activity in vitro, and for piRNA maturation and transposon silencing in vivo. We propose a model for piRNA biogenesis in animal germ lines, in which the Zuc endoribonuclease has a key role in primary piRNA maturation.

Structural basis for nonribosomal peptide synthesis by an aminoacyl-tRNA synthetase paralog.

Cyclodipeptides are secondary metabolites biosynthesized by many bacteria and exhibit a wide array of biological activities. Recently, a new class of small proteins, named cyclodipeptide synthases (CDPS), which are unrelated to the typical nonribosomal peptide synthetases, was shown to generate several cyclodipeptides, using aminoacyl-tRNAs as substrates. The Mycobacterium tuberculosis CDPS, Rv2275, was found to generate cyclodityrosine through the formation of an aminoacyl-enzyme intermediate and to have a structure and oligomeric state similar to those of the class Ic aminoacyl-tRNA synthetases (aaRSs). However, the poor sequence conservation among CDPSs has raised questions about the architecture and catalytic mechanism of the identified homologs. Here we report the crystal structures of Bacillus licheniformis CDPS YvmC-Blic, in the apo form and complexed with substrate mimics, at 1.7-2.4-Å resolutions. The YvmC-Blic structure also exhibits similarity to the class Ic aaRSs catalytic domain. Our mutational analysis confirmed the importance of a set of residues for cyclodileucine formation among the conserved residues localized in the catalytic pocket. Our biochemical data indicated that YvmC-Blic binds tRNA and generates cyclodileucine as a monomer. We were also able to detect the presence of an aminoacyl-enzyme reaction intermediate, but not a dipeptide tRNA intermediate, whose existence was postulated for Rv2275. Instead, our results support a sequential catalytic mechanism for YvmC-Blic, with the successive attachment of two leucine residues on the enzyme via a conserved serine residue. Altogether, our findings suggest that all CDPS enzymes share a common aaRS-like architecture and a catalytic mechanism involving the formation of an enzyme-bound intermediate.

Structural basis for translational fidelity ensured by transfer RNA lysidine synthetase.

Maturation of precursor transfer RNA (pre-tRNA) includes excision of the 5' leader and 3' trailer sequences, removal of introns and addition of the CCA terminus. Nucleotide modifications are incorporated at different stages of tRNA processing, after the RNA molecule adopts the proper conformation. In bacteria, tRNA(Ile2) lysidine synthetase (TilS) modifies cytidine into lysidine (L; 2-lysyl-cytidine) at the first anticodon of tRNA(Ile2) (refs 4-9). This modification switches tRNA(Ile2) from a methionine-specific to an isoleucine-specific tRNA. However, the aminoacylation of tRNA(Ile2) by methionyl-tRNA synthetase (MetRS), before the modification by TilS, might lead to the misincorporation of methionine in response to isoleucine codons. The mechanism used by bacteria to avoid this pitfall is unknown. Here we show that the TilS enzyme specifically recognizes and modifies tRNA(Ile2) in its precursor form, thereby avoiding translation errors. We identified the lysidine modification in pre-tRNA(Ile2) isolated from RNase-E-deficient Escherichia coli and did not detect mature tRNA(Ile2) lacking this modification. Our kinetic analyses revealed that TilS can modify both types of RNA molecule with comparable efficiencies. X-ray crystallography and mutational analyses revealed that TilS specifically recognizes the entire L-shape structure in pre-tRNA(Ile2) through extensive interactions coupled with sequential domain movements. Our results demonstrate how TilS prevents the recognition of tRNA(Ile2) by MetRS and achieves high specificity for its substrate. These two key points form the basis for maintaining the fidelity of isoleucine codon translation in bacteria. Our findings also provide a rationale for the necessity of incorporating specific modifications at the precursor level during tRNA biogenesis.

Decreased aminoacylation in pathology-related mutants of mitochondrial tRNATyr is associated with structural perturbations in tRNA architecture.

A growing number of human pathologies are ascribed to mutations in mitochondrial tRNA genes. Here, we report biochemical investigations on three mt-tRNA(Tyr) molecules with point substitutions associated with diseases. The mutations occur in the atypical T- and D-loops at positions homologous to those involved in the tertiary interaction network of canonical tRNAs. They do not correspond to tyrosine identity positions and likely do not contact the mitochondrial tyrosyl-tRNA synthetase during the aminoacylation process. The impact of these substitutions on mt-tRNA(Tyr) tyrosylation and structure was investigated using the corresponding tRNA transcripts. In vitro tyrosylation efficiency is decreased 600-fold for mutant A22G (mitochondrial gene mutation T5874C), 40-fold for G15A (C5877T), and is without significant effect on U54C (A5843G). Comparative solution probings with lead and nucleases on mutant and wild-type tRNA(Tyr) molecules reveal a greater sensitivity to single-strand specific probes for mutants G15A and A22G. For both transcripts, the mutation triggers a structural destabilization in the D-loop that propagates toward the anticodon arm and thus hinders efficient tyrosylation. Further probing analysis combined with phylogenetic data support the participation of G15 and A22 in the tertiary network of human mt-tRNA(Tyr) via nonclassical Watson-Crick G15-C48 and G13-A22 pairings. In contrast, the pathogenic effect of the tyrosylable mutant U54C, where structure is only marginally affected, has to be sought at another level of the tRNA(Tyr) life cycle.

Crystal structure of human mitochondrial tyrosyl-tRNA synthetase reveals common and idiosyncratic features.

We report the structure of a strictly mitochondrial human synthetase, namely tyrosyl-tRNA synthetase (mt-TyrRS), in complex with an adenylate analog at 2.2 A resolution. The structure is that of an active enzyme deprived of the C-terminal S4-like domain and resembles eubacterial TyrRSs with a canonical tyrosine-binding pocket and adenylate-binding residues typical of class I synthetases. Two bulges at the enzyme surface, not seen in eubacterial TyrRSs, correspond to conserved sequences in mt-TyrRSs. The synthetase electrostatic surface potential differs from that of other TyrRSs, including the human cytoplasmic homolog and the mitochondrial one from Neurospora crassa. The homodimeric human mt-TyrRS shows an asymmetry propagating from the dimer interface toward the two catalytic sites and extremities of each subunit. Mutagenesis of the catalytic domain reveals functional importance of Ser200 in line with an involvement of A73 rather than N1-N72 in tyrosine identity.

Tyrosyl-tRNA synthetase: the first crystallization of a human mitochondrial aminoacyl-tRNA synthetase.

Human mitochondrial tyrosyl-tRNA synthetase and a truncated version with its C-terminal S4-like domain deleted were purified and crystallized. Only the truncated version, which is active in tyrosine activation and Escherichia coli tRNA(Tyr) charging, yielded crystals suitable for structure determination. These tetragonal crystals, belonging to space group P4(3)2(1)2, were obtained in the presence of PEG 4000 as a crystallizing agent and diffracted X-rays to 2.7 A resolution. Complete data sets could be collected and led to structure solution by molecular replacement.

Evolution of the tRNA(Tyr)/TyrRS aminoacylation systems.

The tRNA identity rules ensuring fidelity of translation are globally conserved throughout evolution except for tyrosyl-tRNA synthetases (TyrRSs) that display species-specific tRNA recognition. This discrimination originates from the presence of a conserved identity pair, G1-C72, located at the top of the acceptor stem of tRNA(Tyr) from eubacteria that is invariably replaced by an unusual C1-G72 pair in archaeal and eubacterial tRNA(Tyr). In addition to the key role of pair 1-72 in tyrosylation, discriminator base A73, the anticodon triplet and the large variable region (present in eubacterial tRNA(Tyr) but not found in eukaryal tRNA(Tyr)) contribute to tyrosylation with variable strengths. Crystallographic structures of two tRNA(Tyr)/TyrRS complexes revealed different interaction modes in accordance with the phylum-specificity. Recent functional studies on the human mitochondrial tRNA(Tyr)/TyrRS system indicates strong deviations from the canonical tyrosylation rules. These differences are discussed in the light of the present knowledge on TyrRSs.

Human mitochondrial TyrRS disobeys the tyrosine identity rules.

Human tyrosyl-tRNA synthetase from mitochondria (mt-TyrRS) presents dual sequence features characteristic of eubacterial and archaeal TyrRSs, especially in the region containing amino acids recognizing the N1-N72 tyrosine identity pair. This would imply that human mt-TyrRS has lost the capacity to discriminate between the G1-C72 pair typical of eubacterial and mitochondrial tRNATyr and the reverse pair C1-G72 present in archaeal and eukaryal tRNATyr. This expectation was verified by a functional analysis of wild-type or mutated tRNATyr molecules, showing that mt-TyrRS aminoacylates with similar catalytic efficiency its cognate tRNATyr with G1-C72 and its mutated version with C1-G72. This provides the first example of a TyrRS lacking specificity toward N1-N72 and thus of a TyrRS disobeying the identity rules. Sequence comparisons of mt-TyrRSs across phylogeny suggest that the functional behavior of the human mt-TyrRS is conserved among all vertebrate mt-TyrRSs.

Toward the full set of human mitochondrial aminoacyl-tRNA synthetases: characterization of AspRS and TyrRS.

The human mitochondrion possesses a translational machinery devoted to the synthesis of 13 proteins. While the required tRNAs and rRNAs are produced by transcription of the mitochondrial genome, all other factors needed for protein synthesis are synthesized in the cytosol and imported. This is the case for aminoacyl-tRNA synthetases, the enzymes which esterify their cognate tRNA with the specific amino acid. The genes for the full set of cytosolic aaRSs are well defined, but only nine genes for mitochondrial synthetases are known. Here we describe the genes for human mitochondrial aspartyl- and tyrosyl-tRNA synthetases and the initial characterization of the enzymes. Both belong to the expected class of synthetases, have a dimeric organization, and aminoacylate Escherichia coli tRNAs as well as in vitro transcribed human mitochondrial tRNAs. Genes for the remaining missing synthetases were also found with the exception of glutaminyl-tRNA synthetase. Their sequence analysis confirms and further extends the view that, except for lysyl- and glycyl-tRNA synthetases, human mitochondrial and cytosolic enzymes are coded by two different sets of genes.