The Princeton Guide to Evolution

V.8

Epigenetics

Florian Maderspacher

OUTLINE

1. The concept of epigenetics

2. The history of epigenetics

3. Epigenetics and gene regulation

4. Molecular epigenetics

5. Epigenetic processes

6. Transgenerational epigenetic effects

7. Lamarckism and neo-Lamarckism

8. Epigenetics and evolution

9. Plasticity and assimilation

10. Epilogue

Although coined by Waddington more than 70 years ago, the term epigenetic has become widely used only in the past 15 years. A concept, rather than a discipline, epigenetics is being constantly redefined, often controversially. In its broadest sense, epigenetics refers to stable phenotypic changes without a change in genotype. As phenotypes are the result of gene activity, epigenetics is studied by molecular biologists mainly in the context of gene regulation during cellular differentiation. Of particular interest for evolution are epigenetic changes, often induced by the environment, that can be transmitted across generations. At present, it is unclear whether such epigenetic mechanisms have contributed to evolutionary change.

GLOSSARY

(Genetic) Assimilation. A concept put forward by Conrad Hal Waddington, whereby an externally induced phenotype eventually emerges through random genetic mutations.

Chromatin. The material, DNA and protein, of which eukaryotic chromosomes are made. The DNA is wrapped around complexes of histone proteins, forming so-called nucleosomes, in order to accommodate the long DNA strands in the cell nucleus.

Epialleles. States of a gene with the same nucleotide sequence, but different activity that can be transmitted across generations. Known examples in plants and animals are due to differential DNA methylation.

Germ Line. The sequence of genetic material that gets passed on continually from generation to generation. It is distinguished from the soma, which perishes with each individual generation.

Histones. Nuclear proteins that package DNA into chromatin. Histones are chemically modified in many different ways, for instance, by addition of acetyl, methyl, or phosphate groups.

(Genetic) Imprinting. Silencing of alleles of a gene depending on whether they originate from the father or the mother. Imprinting is found in animals and plants and thought to be the consequence of genetic conflicts between parents.

Paramutation. Interaction between two alleles of a gene, whereby one allele permanently silences expression from the other. Examples are known from maize and possibly mice.

Reprogramming. The resetting of the epigenetic state of a cell or chromosome, either naturally or artificially induced. It occurs regularly in the life cycle of many plants and animals and often involves DNA demethylation.

Soft Inheritance. A term coined by Ernst Mayr, referring to the Lamarckian idea that nongenetic changes in phenotype can be transmitted to the next generations.

Transposable Elements. Mobile genetic units that can “jump” between regions of the genome and make up a large fraction of the genome (40% in humans, 85% in maize).

X-Inactivation. A mechanism for balancing gene dosage in some animals in which one sex has more copies of a sex chromosome than the other. In female mammals, one of the two X chromosomes is randomly silenced in somatic tissues.

1. THE CONCEPT OF EPIGENETICS

Epigenetics is not so much a discipline related to a particular object or process of study but rather a concept, the meaning of which has been (and is being) constantly redefined and debated. Today, the word epigenetic is used mainly in the fields of molecular and evolutionary biology. Molecular biologists use it to refer to stable changes in gene activity that are not due to a change in the DNA sequence itself, or more often, to loosely refer to chemical modifications of DNA and its associated proteins. Evolutionary biologists use it to describe nongenetic, often induced, phenotypic changes that can be transmitted across generations.

A few common themes underlie the various meanings of epigenetic: first, there needs to be a phenotypic effect: a morphological, biochemical, or behavioral alteration in the organism. For evolutionary considerations, it is important whether the phenotypic effect alters the fitness of the organism and whether it is a response to a change in the environment. Second, the effect should be stably transmitted across cell divisions or across generations, and it should in principle be reversible. Third, the effect should not be due to a change in DNA sequence. This last point is the most basic common denominator of epigenetics, as most of its definitions aim at distinguishing epigenetics from genetics.

2. THE HISTORY OF EPIGENETICS

The term epigenetic (literally meaning “above” or “after genetics”) was coined by Conrad Hal Waddington in 1939 to refer to causal changes by which “the genes of the genotype bring about genetic effects.” Waddington was interested in the ways in which genes control embryonic development and chose the term for its similarity to epigenesis—the notion that organisms develop de novo, which was historically set against “preformation,” the (erroneous) idea that the structures of the organism are already fully contained in the egg or sperm. For Waddington, the “epigenetic constitution” comprised the signals that control genes to “bring the phenotype into being.” In 1959, David Nanney, who studied cytoplasmic inheritance in the ciliate protist Tetrahymena, described epigenetic systems as “systems regulating the expression of genetic potentialities.” Nanney also noted that epigenetic changes should be reversible. Later, Joshua Lederberg suggested the term epinucleic information for all information not directly dependent on the DNA sequence itself.

Until about 20 years ago, the adjective epigenetic was rarely used, but in the last 15 years its relative usage in the literature has increased 10-fold compared with the previous 15 years. The modern usage is much narrower than Waddington’s original definition and essentially equates epigenetics with chemical modifications of DNA and its associated proteins. This usage can be traced back to Robin Holliday, who in the 1970s and 1980s postulated that the methylation of DNA could serve as a heritable epigenetic mechanism of gene expression control. In the late 1990s and early 2000s, histones—proteins that package DNA in the nuclei of eukaryotic cells—were found to be chemically modified in many different ways. These modifications were interpreted by some as a “histone code” that carries epigenetic information about gene regulation. Around the same time, a renewed interest in the evolutionary aspects of epigenetics emerged. In particular, Eva Jablonka has promoted the idea that instances of acquired traits that are inherited across generations call for an overhaul of Darwinian evolution.

3. EPIGENETICS AND GENE REGULATION

What “brings the phenotype into being” are the gene products—RNAs and proteins—expressed by a given organism or cell. A nerve cell is different from a muscle cell, because it expresses different genes, although both cell types share the same DNA sequence; thus, the difference between the two is epigenetic. Whether a given gene is activated or not is decided mainly at the level of transcription; that is, whether or not a messenger RNA is copied from the gene, and in what quantity. This decision is computed by transcription factors that either promote the gene’s transcription or repress it and bind in a sequence-specific manner to regulatory DNA elements located around the gene (see chapter V.7). Small RNA molecules also contribute to gene regulation by preventing messenger RNAs from being translated into proteins. Whether a gene is active or not thus depends on the regulatory sequences associated with it, a genetic component, and the presence of the transcriptional and translational regulators, an epigenetic component.

The transcription factors themselves are regulated in response to internal signals, such as hormones or growth factors. Environmental cues, for instance, a change in temperature, can also affect gene activity. During the development of an organism, changes in gene expression are caused by signals that are often transient in nature, yet whose effect on gene expression needs to be stable, such that a muscle cell always remains a muscle cell. Cells thus form a kind of epigenetic memory of gene activity.

The simplest way of maintaining a gene active or inactive in the absence of the initial regulatory signal is through feedback loops of transcription factors that regulate each other’s expression. For instance, the mammalian muscle differentiation factor myoD activates the transcription of several other transcription factors that in turn maintain myoD transcription. These factors are passed on during cell divisions and ensure continued activity of the muscle program in daughter cells. Although such mechanisms are found abundantly during embryonic development and fit Waddington’s original definition, they are nowadays rarely referred to as “epigenetic.” At the same time, the relationship of what is now called “epigenetic” to gene regulatory mechanisms is a matter of active research and debate.

4. MOLECULAR EPIGENETICS

Modification of DNA-Associated Proteins

In eukaryotic cells, the genetic material is stored as chromosomes, which consist of chromatin—DNA wrapped around protein complexes of histone proteins. Chromatin is packaged to accommodate the length of the DNA strands in the cell nucleus. It is thought that the degree of packaging is closely correlated with the transcriptional activity of a particular DNA region: active regions are loosely packed such that the transcriptional machinery can gain access, while inactive regions are packed tightly. The degree of packaging depends on various chemical modifications attached to the histones at defined sites.

There are dozens of different histone modifications that—depending on their position and chemical nature—can have different effects on chromatin packaging and other biological functions. The association of a given modification with a particular function or activity state of a gene has led to the idea of a “histone code” or “epigenetic code.” Acetyl groups on histones, for instance, are generally associated with transcriptional activity, while other modifications, such as methyl groups, can have different effects depending on where they are added. It is thus likely that the function of an individual histone mark depends on the overall context, rather than being defined by a rigid code.

Two important questions concern the possible epigenetic function of histone modifications: First, how exactly do histone modifications relate to gene activity? There is much evidence that histone modifications can be influenced by physiological or pathological changes in the cell as well as by the environment, but to what extent these changes are a cause or an effect of altered gene expression is not clear. Second, are histone modifications stable, and can they be faithfully transmitted across cell divisions? While there are some indications that histone marks can be replicated in the absence of the original inducers, a clear replication mechanism has yet to be found. And even if there is transmission across cell divisions, is it self-perpetuating, or maintained by trans-acting transcription factors? Most evidence up to now points toward histone marks as a consequence rather than a cause of changes in gene expression effected by the transcriptional machinery.

A particular class of proteins is associated with preserving a cell’s memory of gene activity: the polycomb group (PcG) and trithorax group (trxG) proteins. Broadly speaking, PcG proteins keep silent genes silent, while trxG proteins keep active genes active even after the initial regulators have disappeared. Most of the genes targeted by PcG and trxG proteins encode transcriptional regulators themselves, in particular the Hox genes that define the identity of body regions in animals (see chapter V.11). In Drosophila melanogaster, the decision about which Hox gene is active where is made early in development, but their inducers are present for only a short time. The activity status of a given Hox gene is preserved by PcG and trxG proteins, such that the tissue in an adult fly will retain the Hox identity of its embryonic precursor

DNA Methylation

Of all the putative carriers of epigenetic information, DNA methylation has received most attention. DNA’s cytosine bases can have a methyl group attached, giving rise to 5-methylcytosine. Aspects of the methylation machinery are similar between divergent types of organisms, suggesting that DNA methylation evolved very early in the history of life. Methylated DNA is generally associated with inactivity of genes, such that when methylation is lost, previously silent genes become transcribed. Aberrant DNA methylation is, for instance, associated with irregular gene activity found in some cancers. Methylation serves mainly to keep genes silent in the long term, but it is a consequence rather than a cause of inactivated gene expression. DNA methylation is particularly important for the silencing of transposable elements. Transposable elements can be potentially harmful if mobilized; thus protection against transposable elements and related genetic parasites may be a primary function of DNA methylation. DNA methylation can be inherited across cell divisions through a mechanism closely linked to the replication of the DNA strands themselves; however, the fidelity with which DNA methylation marks are inherited is many orders of magnitude lower than that of DNA replication, and they generally do not persist across generations.

The extent of DNA methylation varies greatly between species: yeast (Saccharomyces cerevisiae) and the nematode Caenorhabditis elegans show no DNA methylation, Drosophila fruit flies have very little, while maize and mammals show extensive methylation. Likewise, the way DNA methylation is distributed across the genome varies greatly: the genomes of mammals are methylated throughout, with the exception of the regulatory regions of many genes. In some plants, like maize, the genome is almost completely methylated, while in the plant model Arabidopsis thaliana, methylation is clustered around repetitive DNA and transposable elements. The degree to which DNA methylation patterns vary between individuals and over time is currently an active area of research. In Arabidopsis, for instance, DNA methylation patterns vary between plants that are genetically nearly identical. Most importantly, DNA methylation is relatively stable over generations but in some instances switches back and forth. Much like changes in gene expression, methylation patterns can also vary with age and between different environments.

The predominant function of DNA methylation (and to some extent histone modifications) is to keep genes whose functions in a given cell are not needed—or in the case of transposable elements, not wanted—reliably silenced. Molecular epigenetic systems are thus, quite in the sense of Waddington, vital for bringing the phenotype into being, but they can act only on the genetic program encoded in the DNA sequence. And of course, the whole molecular epigenetic machinery is itself encoded in the genes.

5. EPIGENETIC PROCESSES

The best-studied epigenetic processes concern the permanent silencing of alleles or entire chromosomes. Genomic imprinting, for example, is an epigenetic process in animals and plants that leads to the expression of certain genes in the offspring from only one of the two alleles—either the one inherited from the father, or the one from the mother. In mammals about 100 genes are imprinted, most of which are involved in mother-offspring interactions. Imprinting is established in the germ cells according to the sex of the organism, and in mammals is initiated by DNA binding proteins or noncoding RNAs. In both animals and plants, all the epigenetic systems mentioned in the previous section are involved in maintaining the silenced state.

Genomic imprinting has been proposed to be the evolutionary consequence of a “conflict of interest” between the genes of the mother and those of the father. In placental mammals, where the embryo is nourished by the mother, it is in the mother’s interest not only to feed the embryo, but also to protect her own resources, while the father’s genes benefit only from a thriving embryo. In line with this, most of the paternally imprinted genes promote the growth of the embryo, while those expressed from the maternal allele restrict it.

During X-inactivation, one of the two X chromosomes is permanently silenced in the cells of female mammals. This ensures the same amount of X-derived gene product is made as in males, which have only one X chromosome. X-inactivation is initiated by a noncoding RNA (Xist) that marks the chromosome to be silenced, and DNA methylation is necessary to keep the silenced X chromosome silent (see chapter V.4).

Another paradigmatic epigenetic process is paramutation, which has been best studied in the maize b1 gene that promotes purple pigmentation. Two alleles differentially affect b1 transcription but share an identical DNA sequence: the B-I allele shows high b1 expression, while the B’ allele shows low expression. When the two alleles are present in the same plant, B’ will suppress the expression of b1 from the B-I allele, such that the heterozygous plant looks like a B’ plant. The B-I allele is said to be paramutated and will functionally behave like a B’ allele in subsequent generations; moreover, it will be able to paramutate other B-I alleles, an effect that lasts for many generations.

Although epigenetic marks, such as those established during X-inactivation, can be stable in somatic cells, the marks are erased in many organisms before the beginning of a new generation, a process known as reprogramming. In the mammalian life cycle, there are two major episodes of reprogramming: in the early embryo after fertilization, and in the primordial germ cells of embryos. Reprogramming is important in various contexts, especially so that imprints on genes can be reset and renewed according to the sex of the embryo, but also for the fertilized eggs to become pluripotent; that is, to be able to give rise to all the different cell types of the organism. Scientists have exploited this particular aspect of reprogramming to generate induced pluripotent stem (iPS) cells from differentiated somatic cells of mammals by adding extraneous transcriptional regulators that revert the differentiated state of a cell.

Unlike animals, plants do not show a global reprogramming in germ cells but rather in the tissues that nourish the plant embryo. From there, small RNAs are thought to reinforce silencing of transposable elements in the embryo. A classic example of a reprogrammed epigenetic response in plants is vernalization. Many plants in temperate regions require a prolonged exposure to cold to induce fast flowering and seed development. Individual plants form a long-lasting epigenetic memory of the transient cold experience. In Arabidopsis, the vernalization response is associated with silencing of the gene FLC, which encodes a repressor of flowering. In seeds, FLC silencing is reset, which is why the offspring of vernalized plants do not retain a memory of the vernalization response.

6. TRANSGENERATIONAL EPIGENETIC EFFECTS

Of particular interest to evolutionary questions are epigenetic changes that are heritable across generations, for which there are a handful of examples in plants and animals. These can come in the form of more or less stable epialleles, which involve gene silencing associated with altered DNA methylation, or they can be induced through environmental stimuli.

Plants

Plants differ from animals in two ways that are relevant for epigenetics: first, they do not set aside their germ line early in development, which along with the absence of large-scale reprogramming might make them more prone to carrying epigenetic changes into the next generation. Second, plants cannot move and thus may have a greater need to modulate their phenotype in response to the environment. Because seed dispersal is often limited, offspring will grow up in an environment similar to that of their parents; it thus may be beneficial for them to be prepared for that environment through a kind of transgenerational epigenetic memory.

Three particularly well-characterized examples of transgenerational epigenetic effects come from heritable epialleles in plants. In toadflax (Linaria vulgaris), so-called peloric mutants have radially symmetrical flowers instead of asymmetrical ones. In tomatoes (Solanum lycopersicum), colorless nonripening mutants do not turn red, and their flesh disintegrates instead of softening. In melons (Cucumis melo), certain gynoecious lines, unlike other strains, only form female flowers. In all these cases, the phenotype is relatively stable across generations and has been linked to lowered expression at the underlying gene without a detectable difference in nucleotide sequence between mutant and normal plants; instead, all these mutants show excessive DNA methylation near the silenced gene. However, an adaptive value is not apparent in any of these cases.

Plants have evolved a number of phenotypic responses to environmental stress or predators, and in some instances such responses have been shown to be heritable. One example for such an induced transgenerational effect comes from the yellow monkeyflower (Mimulus guttatus). In some strains of Mimulus, the density of trichomes—hairs that serve as a defense against insect herbivores—increases in response to insect damage. Offspring from mother plants that showed this response will form more trichomes, even if they have experienced no leaf damage themselves; thus the induced phenotype is transmitted from mother to offspring. While a possible adaptive value of this response is conceivable, it is not clear how it is transmitted, nor for how many generations the effect lasts.

In Arabidopsis, stressors, such as UV light or bacterial proteins, induce genome instability—a so-called genomic shock. This effect is heritable through both the maternal and paternal germ lines and persists for about four generations in offspring that do not experience the stressors themselves. The precise basis for this effect is not clear, nor is its possible adaptive value, but DNA methylation and small RNAs are necessary for its transmission. A similar transgenerational effect with a possible advantageous function was found for temperature. When Arabidopsis plants were exposed to heat, their third-generation offspring, when confronted with a hot environment, produced many more seeds than plants whose ancestors had not experienced heat.

Animals

In animals, the distinction between germ line and soma is made early in development, and while epigenetic mechanisms play a role in gene silencing during somatic differentiation, the extent to which environmentally induced effects can become transmitted through the germ line is not clear. When transgenerational effects in mammals are considered, it is worth bearing in mind that in the mammalian germ line epigenetic marks are widely reset. Moreover, a pregnant female contains not only her own offspring but also the germ cells of that offspring; thus, an environmental effect on the mother can in principle directly affect the two following generations without necessarily implying transgenerational epigenetic inheritance.

The best-studied transgenerational effect in animals comes from an epiallele of the mouse coat color gene agouti, which encodes a protein regulating pigment synthesis. Offspring of mothers carrying the Agouti viable yellow (Avy) allele can range from yellow to nearly normal coat color despite being genetically nearly identical. The range of phenotypes will depend on the phenotype of the mother: darker mothers tend to have more dark offspring, while lighter mothers have more light offspring. The Avy allele is caused by a transposable element near the agouti gene, and Avy alleles differ in their degree of DNA methylation: highly methylated alleles lead to less agouti expression and darker offspring, whereas less methylated alleles lead to yellow offspring. Most likely, the epiallele persists through reprogramming because of the transposable element being targeted by silencing mechanisms. Interestingly, genetic alleles of agouti have been implicated in evolutionary adaptation of coat color in beach mice (see chapter V.12).

Three recent examples of induced transgenerational epigenetic effects come from the nematode worm C. elegans, a popular model organism. In C. elegans, an introduced RNA virus will trigger a silencing response that prevents expression of the virus components through small RNAs that are complementary to viral sequences. This response can persist over several generations, in some individuals indefinitely, even if the virus is removed and reintroduced in subsequent generations. While this response may be of obvious adaptive value to the worms, it is not clear whether it is relevant in the context of naturally occurring viruses, as the virus was introduced artificially. In another example, the offspring of C. elegans worms reared in the presence of attractive odors retain an increased preference for these odors even if they have not experienced them before. If raised in the same conditions over more than four generations, offspring can retain the olfactory imprint for at least 40 generations. This transgenerational imprinting has the potential to prime a new generation’s preferences for the presumably favorable conditions in which its ancestors grew up. The third example concerns life span. Genetic mutants for several components of the machinery responsible for generating certain histone modifications live longer by about 20–30 percent. When from these mutants offspring are generated that no longer carry the mutant alleles, these worms still live longer than worms whose ancestors did not carry the mutations. What causes this effect, which vanishes after three or four generations, is not known.

A very well-studied transgenerational effect in mammals concerns maternal care behaviors of rats. When female rats are stressed, they dedicate less time to maternal care. Pups that receive less maternal care will in turn provide less maternal care to their own pups. Less maternal care ultimately translates into lower glucocorticoid receptor expression during the first week of a pup’s life. This change in gene activity is carried over into adulthood—when the stimulus provided by maternal care is no longer present—possibly through increases in DNA methylation and histone modifications. Interestingly, this effect is seen not only for a mother’s natural, genetically related offspring but also when unrelated pups are introduced to the nest; therefore, the effect must be transmitted completely independently of the germ line, through the mother’s behavior.

Another transgenerational effect in rats is due to chemicals that interfere with the hormonal regulation of sexual development. Compounds such as the fungicide vinclozolin can interfere with steroid signaling during gonad development. Male rats exposed in utero to vinclozolin show reproductive defects, most notably reduced sperm number and motility. These effects on male fertility are transmitted across several generations in a sex-specific manner: offspring will inherit the effect from their fathers but not from their mothers.

Humans

Unless one chooses to include culturally transmitted traits such as language—which some do—the evidence for transgenerational epigenetic effects in humans is scant. The most popularized cases concern the effects of parental malnutrition on offspring health. For instance, the children of Dutch women who were pregnant during a famine in World War II were found to be more likely to develop type 2 diabetes; however, this particular effect lasted only into the first generation and thus may simply have been a maternal effect of exposure in the uterus. The only evidence for an effect extending into the second generation comes from the Swedish Överkalix population. In the nineteenth century, increased food availability for males between ages 8 and 12 led to a shorter life span in their grandsons as a result of cardiovascular disease and diabetes. A similar relationship was also seen for grandmothers and their granddaughters. However, such effects have so far not been described elsewhere; thus confounding environmental or social factors cannot easily be ruled out.

7. LAMARCKISM AND NEO-LAMARCKISM

The notion that traits that change as a response to the environment could be inherited is associated with Jean Baptiste Lamarck. Based on the well-known malleability of traits—the muscles of the blacksmith grow as he works—Lamarck posited in 1809 that changed environments lead to changed habits in the use of an organ (see chapter I.2). Through inheritance of the acquired state of the organ, the organ will increase or decrease in the offspring. Lamarck’s theory initially received little recognition, but similar ideas were widespread, and even Darwin incorporated them into his theory. Only in the last third of the nineteenth century did Lamarckian ideas gain popularity as an alternative to natural selection, which was rejected by many on moral grounds.

At the same time, experimental evidence was mounting against the inheritance of acquired traits. August Weismann conducted a famous experiment in which he cut off the tails of mice in five subsequent generations and in nearly a thousand offspring never observed any tail shortening. Weismann rejected the idea of inheritance of acquired traits because it was incompatible with his idea of germplasm—what we would today call the genome—that was propagated independently and separate from the somatic cells. In addition, Theodor Boveri’s demonstration that the chromosomes determine the traits of a cell, the rediscovery of Gregor Mendel’s laws of inheritance, and finally the chromosome theory of inheritance by Thomas Hunt Morgan established that genetic factors controlled the phenotype. This notion became enshrined in evolutionary thinking through the modern synthesis that unified genetics and Darwinian evolution (see chapter I.2).

Lamarckian ideas became further discredited because of two men, Paul Kammerer and Trofim Lysenko. Kammerer had claimed that breeding in water induced mating pads in midwife toads. Midwife toads usually lack these pads, as they breed on land, but because he found that the induced pads could be inherited, Kammerer publicized this as an instance of Lamarckian inheritance. Later, he was accused of fraud, and ended his life. Lysenko studied the vernalization response in crops and claimed it too could be inherited, and that in fact all inheritance was due to acquired characteristics—an idea that resonated well with Communist ideology. Lysenko rose quickly in the ranks of Stalin’s Russia and was responsible for widespread agricultural malpractice.

With the advances in molecular epigenetics, a renewed interest in Lamarckian inheritance, sometimes called soft inheritance, has emerged in some circles. This interest is due in part to claims that DNA sequence variation alone was not able to explain the diversity of life or the notion of “missing heritability” in human diseases—despite many diseases being heritable, the individual DNA sequence variants found associated with a disease contribute statistically very little to an individual’s disease risk (see chapter V.13). To some extent, this interest is echoed in the public domain and can be seen as a backlash to overblown claims of genetic determinism at the onset of the genomic age.

8. EPIGENETICS AND EVOLUTION

The question whether and to what extent epigenetic processes contribute to evolution is a matter of debate. Compared with the vast number of genetically transmitted traits, transgenerational epigenetic effects are indeed extremely rare (or very difficult to detect). Many of the best-documented cases of phenotypic change without concomitant genetic change are epialleles. Epialleles change the activity status of a gene, without changing its DNA sequence, but they are most likely independent of environmental inducers. In the cases known so far, the “mutant” epiallele causes an increase in DNA methylation, thus leading to reduced function of the gene. In many cases, epialleles are linked to transposable elements that remain stably methylated, presumably as part of a genomic defense. From an evolutionary point of view, such epialleles behave largely like genetic alleles. Their frequency could in principle increase or decrease in a population as the result of natural selection depending on their effect on fitness; however, epialleles can be more variable than genetic alleles, as the case of agouti illustrates. They could thus provide an additional source of variation, which might be advantageous in fluctuating environments; although for most known epialleles, an adaptive value is not readily apparent. Epialleles are also less stable and revert more frequently to the alternative state, which means that in each generation some of the variation may be lost. So far, there is no compelling evidence that epialleles have been the target of natural selection.

By contrast, heritable epigenetic changes that are induced as a response to environmental stimuli behave rather differently from genetic alleles. Genetic mutations arise more or less randomly (though some regions of the genome are more prone to mutations than others) and independent of the selective pressures and phenotypic change they may effect. New mutations will be rare, and even if they have a beneficial effect, they may either be lost from the population by genetic drift, or they can take several generations to become frequent in the population. By contrast, an induced plastic response can in principle affect many members of a population at once. When the change is heritable, it thus has the potential to lead to widespread adaptation much faster than genetic change. Hence, heritable induced epigenetic effects have been postulated to operate on an intermediate timescale, between the short-term phenotypic plasticity that affects only one generation and the long-term genetic adaptation that takes multiple generations to take effect. However, so far no case is known in which an induced heritable response has been shown to underlie an evolutionary change in nature.

9. PLASTICITY AND ASSIMILATION

The notion of the inheritance of acquired traits is closely linked to phenotypic plasticity—the ability of a given genotype to generate different phenotypes depending on environmental conditions (see chapter V.11). Phenotypic plasticity is a basic property of organisms and vital for coping with changing environments. If the environmental fluctuations last longer than the generation time of an organism, it may be advantageous if the offspring is already endowed with the adaptive phenotype. One way of carrying over a plastic response into the next generation is through maternal effects, whereby the mother influences the phenotype of her offspring, for example, by adjusting the chemical composition of the egg; this need not necessarily invoke a hereditary mechanism in the strong (neo-)Lamarckian sense.

An acquired trait that is transmitted over several generations (or is induced anew in each generation) might eventually become fixed genetically if mutations that cause the same phenotype as the induced response accumulate in its carriers. This effect was postulated by Waddington as “genetic assimilation” and tested experimentally. While these ideas have been widely popularized, because of their appeal for explaining fast adaptation, there is currently no direct evidence for either genetic assimilation or the inheritance of acquired traits playing a major role in evolutionary change. By contrast, there are by now numerous examples for the genetic basis of evolutionary adaptations. It is also worth keeping in mind that the very ability to respond to an environmental stimulus with a meaningful change in phenotype requires a dedicated machinery that translates the environmental stimulus into a change in gene expression—a machinery that of course must be genetically encoded and evolved by natural selection.

10. EPILOGUE

Epigenetics in its many guises touches on various ideas that exert great attraction to researchers and the public alike: for the public, because the idea that “your genes are your destiny” has some disconcerting implications if interpreted wrongly; for researchers, because there is a temptation to discover new phenomena that challenge or overthrow dominant paradigms, but also because transgenerational epigenetic effects are inherently interesting biological phenomena. This is not so much because they force us to change the way we think about genetics and evolution but because they are the few exceptions to the rule; as the geneticist William Bateson advised biologists, “Treasure your exceptions.”