![](https://blogger.googleusercontent.com/img/b/R29vZ2xl/AVvXsEgd5Ll4TDwOJQn3udljeqnjrS6yb6BwGqL_DoxCL5zn2e14_Uas8Sr9dlVTzs-cSvmEGD9zJ5zyR2ISvNrbHW04YNOtIwivI0fJCihD8VmkZx_3m8FXtjDz88xZw3szrDxzuBGf0V1kmWA/s320/epi1.jpg)
Waddington introduced two important concepts “canalization,” which allows cells with identical genomes to take diverse trajectories and the “epigenetic landscape” that is formed through this process. This original model has influenced our understanding of epigenetics as innate processes that result in canalized terminal differentiation that to a large extent is irreversible.
The concept of epigenetics implies that nongenetic events could generate stable phenotypic differences. Therefore, toxic agents that are not mutagenic could cause stable adverse phenotypic changes if they interfered with epigenetic processes. The classic understanding that epigenetic processes are exclusively involved in embryonal development as proposed by Waddington would imply that toxic agents that interfered with epigenetic mechanisms would affect the phenotype only during gestation. One of the most elegant illustrations that interference of nongenotoxic agents in epigenetic processes during gestation would result in stable phenotypic changes was the demonstration in the agouti (A(vy)) mouse model that maternal dietary methyl content supplementation affected the coat color of her offspring through DNA methylation changes This study provided evidence that a phenotype of an organism could be stably changed by exposure to a nongenotoxic agent during gestation.
However, the cardinal question is how terminal canalization is or how “high” are the walls of the canals that delineate the epigenetic landscape. Do epigenetic processes play a role in altering and modifying the phenotype beyond embryonal development? If indeed the epigenetic landscape is reversible after birth, then it is possible that toxic epigenetic agents influence the phenotype not only during gestation but also later in life.
The mysterious epigenetic processes proposed by Waddington are now understood in biochemical terms and allow new perspective on how nongenotoxic agents could trigger adverse phenotypic changes. There are several epigenetic mechanisms that are intensively investigated, and these include chromatin structure and histone modification that gate the access of transcriptional machinery to genes, noncoding RNAs including microRNA that regulates gene expression through altering chromatin configuration, inhibition of translation, and degradation of RNA , and remarkably, the DNA molecule itself bears epigenetic information encoded in the DNA methylation pattern , which will be the focus of our discussion.
DNA methylation is a covalent modification of DNA by addition of methyl residues to cytosine or adenine bases in DNA. DNA methylation in all organisms targets specific sequences. In vertebrates, the CG dinucleotide sequence is a principal target of DNA methylation because it is preferentially recognized by vertebrate DNA methyltransferases (DNMT). CG is the only dinucleotide sequence that contains a cytosine that is a palindrome and could be copied during cell division by a semiconservative DNMT from the parental strand onto the daughter strand. Thus, changes introduced into the DNA methylation pattern either stochastically or as an organized response to developmental or environmental signals could be maintained and memorized through DNA replication cycles. This is a mechanism through which a transient exposure to an environmental agent could result in lasting impact on DNA methylation and as a consequence on the phenotype. However, recent data including genomic sequencing suggest that DNA methylation occurs in other dinucleotide sequences in addition to CG in undifferentiated cells. It remains to be seen, however, whether non-CG methylation is present, albeit at lower level, in other differentiated tissues such as the brain and whether it plays a role in dynamic DNA methylation responses throughout life. The presence of non-CG methylation in the genome suggests that at least certain methylation marks are not automatically mitotically heritable as predicted by the classical model of semiconservative mitotic heritability of DNA methylation as will be discussed below and are maintained dynamically by a balance of methylating and demethylating enzymes. However, DNA methylation changes in both CG and non-CG sites could potentially mediate the long-term impact of exposure to environmental agents, although there might be different mechanisms of maintenance of these different methylated sequences.
What distinguishes DNA methylation in vertebrates is that not all potential methylatable sequences are methylated in all cells in a given individual. The same site might be methylated in several cell types but not others. This creates a cell type–specific pattern of methylation. Cell type–specific patterns of methylation or tissue-specific differentially methylated regions were discovered in the early 80’s using methylation-sensitive restriction enzymes and were confirmed three decades later by high-throughput genomic sequencing. Thus, in addition to the individual identity that is encoded in the sequence of the four bases in DNA, there is a cell type identity encoded in the distribution of methyl moieties in the same molecule of DNA. Cell type–specific differentially methylated regions provide an elegant explanation for the question of how could the same genome encodes multiple stable phenotypes in a multicellular organism. DNA methylation provides cell type identity to genomes, identical DNA sequence could bear different DNA methylation patterns in different cell types. Thus, alternations in DNA methylation triggered by a toxic environmental agent could have tissue-specific manifestation. This adds a level of complexity to the screening for DNA methylation–modifying agents as the agents might have a diverse impact on the DNA methylation pattern and the gene expression profile in different tissues.