Mediating and maintaining methylation while minimizing mutation: Recent advances on mammalian DNA methyltransferases

Abstract

Mammalian genomes are methylated on carbon-5 of many cytosines, mostly in CpG dinucleotides. Methylation patterns are maintained during mitosis via DNMT1, and regulatory factors involved in processes that include histone modifications. Methylation in a sequence longer than CpG can influence the binding of sequence-specific transcription factors, thus affecting gene expression. 5-Methylcytosine deamination results in C-to-T transition. While some mutations are beneficial, most are not; so boosting C-to-T transitions can be dangerous. Given the role of DNMT3A in establishing de novo DNA methylation during development, it is this CpG methylation and deamination that provide the major mutagenic impetus in the DNMT3A gene itself, including the R882H dominant-negative substitution associated with two diseases: germline mutations in DNMT3A overgrowth syndrome, and somatic mutations in clonal hematopoiesis that can initiate acute myeloid leukemia. We discuss recent developments in therapeutics targeting DNMT1, the role of noncatalytic isoform DNMT3B3 in regulating de novo methylation by DNMT3A, and structural characterization of DNMT3A in various configurations.

Figure 1. Schematic diagram of de novo and maintenance DNA methylation at CpG dinucleotide.

DNMT3A and DNMT3B establish the initial cytosine methylation pattern de novo, together with DNMT3L in germline and DNMT3B3 (or DNMT3B6, not shown) in somatic cells. Open lollipops represent unmethylated C, while filled ones represent 5mC. DNMT1 activity at replication foci is regulated at multiple levels to assure the faithful inheritance of genomic methylation patterns. Factors included in this regulation include PCNA, UHRF1, DNA ligase 1 (LIG1), protein lysine methyltransferases G9a/GLP, among others.

Figure 2. DNMT1-selective, non-nucleoside, dicyanopyridine containing inhibitors.

(a) The chemical structure of GSK3685032. The inhibitors all share a 3,5-dicyanopyridine core but vary in their chemical architecture at positions C2 and C6 of the pyridine ring. (b) Surface representation of inhibitor-bound DNMT1-DNA structure with the inhibitor shown in yellow and the DNMT1 active-site loop shown in magenta. (c) The open conformation of the DNMT1 active-site loop (magenta) in the absence of DNA. Sinefungin is shown in stick model, M, Met1232 and C, Cys1226. (d) The closed conformation of the DNMT1 active-site loop bound to DNA (in orange ribbon; the DNA helical axis is perpendicular to the page). The flipped-out cytosine analog zebularine and SAM are shown in stick model. (e) The active-site Cys1226 and SAM approach the zebularine [83] ring carbon atoms C6 and C5 from opposite directions, respectively. (f) The inhibitor-induced open conformation of the DNMT1 active-site loop bound to DNA. The intra-helical cytosine analog zebularine (cyan) and inhibitor (colored yellow) are shown in stick model. (g) The inhibitor penetrates deeply into the DNA and interacts with DNMT1 located on the major groove side. (h) Sequence alignment of active-site loops among DNMT1, DNMT3A and DNTM3B. The cysteine in the third position (C1226 in DNMT1) is the active site nucleophile.

Figure 3. The top 10 most commonly-observed single-nucleotide mutations in DNMT3A

(derived from a study from ~50,000 healthy individuals [30]). (a) The C-to-T alterations (in red letters) within CpG sites are indicated for mutated codons. The affected amino acids are depicted in the domain structures of DNMT3A. (b) R320 and R326 are located within the PWWP (Pro-Trp-Trp-Pro) domain. (c) R598 is located in the ADD Zn-binding domain conserved in ATRX, DNMT3A and DNMT3L. Three Zn atoms are shown in silver balls. (d) The catalytic domain of DNMT3A (R882H)-DNMT3L tetramer in complex with DNA. (e) R326 forms a salt bridge with D286. (f) Viewing from the point of DNMT3L: R729, R736 and R771 of DNMT3A are located in the interface with DNMT3L. (g) P904 of DNMT3A is part of a curved helix. (h) R882H is located in the DNMT3A dimer interface and interacts with a DNA phosphate.

Figure 4. Structures of DNMT3A in various configurations.

(a) Representatives of catalytically-active members of the DNMT3 family (3A1, 3A2, 3B1 and 3B2) and inactive members (3L, 3B3 and 3B6). For a sequence comparison see figure S1B of [50]. (b) Structure of DNMT3B3–3A-3A-3B3 in complex with nucleosome core particle. DNA is in green, and histones are in gray. (c) DNMT3L-3A-3A-3L in complex with naked DNA. (d) DNMT3L-3B-3B-3L in complex with naked DNA. We note that the structural examples shown here are only for the C-terminal catalytic domain of 3A and 3B, or the equivalent domains of 3L and 3B3. (e) Flanking sequence preferences of DNMT3A and DNMT3B catalytic domains alone. (f) Examples of favored and disfavored DNMT1 methylation sites. (g) Methylation at an Arg codon. The deamination product of 5mC creates a G:T mismatch. Binding of the mismatch by transcription factors limits access to the damaged site and, if not repaired in time, a round of DNA replication converts a CpG dinucleotide to CpA/TpG.