Printer Friendly

Epigenetic Inheritance and Evolution: The Lamarckian Dimension.

Epigenetic changes are heritable variants that are not due to changes in DNA sequence. While these are often discussed in an almost mystical fashion, all organisms display epigenetic variation as a simple consequence of regulated gene expression. Consider the lambda phage, which exists in its bacterial host (Escherichia coli) as either an inactive lysogenic form or as an active lytic form. These states are heritable in that a lysogenic phage gives rise to lysogenic progeny when the bacteria cell divides. The difference between states is not due to variation in DNA sequence, but rather variation in which regulatory proteins are bound to the control regions of the lambda DNA. Once bound, these protein complexes tend to ensure that future lineages retain the parental epigenetic form (the gory details, which have been worked out with exquisite precision, can be found in any molecular biology text). Hence, the lytic and lysogenic states represent different epigenetic states of an identical DNA sequence. The same alternate regulation of identical DNA sequences holds on a far grander scale in multicellular organisms where the collection of differentiated cells that constitute the individual all contain the same DNA sequence. Indeed, multicellularity requires epigenetic mechanisms.

Epigenetics is thus fundamental to understanding both development and gene expression, and not surprisingly, evolutionary biologists have long been fascinated with its proper place in evolutionary theory. Given the impressive progress in our understanding of the mechanisms of gene regulation, it is certainly timely to consider, once again, the evolutionary implications and consequences of epigenetic effects. Enter Jablonka and Lamb, who provide a thoughtful review of the recent molecular literature and suggest a number of potential consequences. In brief, their conclusion is that:

"The new information about the flexibility of the genome has so changed ideas about the nature of the gene and its stability, that several biologists have argued that it is time to look again at some of our basic assumptions about the origin of genetic variation and the role of new variation in evolutionary processes. . . . The present [neoDarwinian] theory is based largely on the assumption that heritable variation is random and involves changes in DNA sequences. If, as we have argued, some variation is not random and is not based on sequence change, but rather is 'acquired' variation induced by the environment, this must alter many parts of evolutionary theory."

Indeed, the authors suggest that not only is Lamarckian inheritance possible, but that epigenetic mechanisms allow it to be placed in a firm molecular framework. A provocative statement indeed! The suggestion that molecular biology offers routes for the inheritance of acquired characters has been made before. Steele et al. (1984) suggested that the enzyme reverse transcriptase (of the AIDS virus, and other viruses) can reverse-transcribe mRNAs from somatic cells into DNA, which is then carried to, and inserted into, germline cells by retroviral vectors. If correct, this pathway allows adaptive mutations culled by somatic selection to be passed onto the next (sexual) generation. Jablonka and Lamb, joining a host of others, note that there are very serious problems with Steele's mechanism. For example, not only is a highly improbable sequence of events required (reverse transcription of the adaptive region, transport to germline cells, insertion), this mechanism is restricted only to those sequences which are transcribed. However, it is likely that much of the genetic variation resides in nontranscribed control regions which would be missed.

Jablonka and Lamb, calling upon new discoveries in molecular biology, suggest that evolutionary biologists have failed to realize that there is "more to inheritance than DNA" and that alternative routes of "inheritance," namely through the stable inheritance of epigenetic variants, allow for Lamarckian evolution. The authors' arguments are thoughtful and based on interpretations of existing data and extrapolations (some mild, others extreme) of known molecular mechanisms. The result is an intriguing book certainly worth reading, but one that in the end has serious, if not fatal, flaws.

Jablonka and Lamb's argument proceeds as follows. First, they suggest that most epigenetic effects arise from differential marking of the DNA, through direct modifications of the DNA such as methylation, through binding of specific regulatory proteins, or through general changes in chromatin structure. While this assumption is not controversial, the remaining postulates are. Building on marking, they argue that (1) environmental effects can induce stable epigenetic changes, (2) in some cases such changes can be maintained even after passing through meiosis, and finally, (3) epigenetic changes, since they involve changes in chromatin structure, may in turn result in differences in the pattern and rate of mutations between sequences in different epigenetic states, potentially allowing an initial epigenetic change to be genetically assimilated as a change in the DNA sequence itself. The authors argue that changes in repetitive DNA sequences are often involved in first predisposing a region to allow for alternate epigenetic states and, further, that changes in repetitive sequences ultimately allow an epigenetic change to eventually manifest itself as a change in DNA sequence variation. We consider these assumptions in turn. While there are problems with all three key arguments, the second is the most suspect because the authors have not demonstrated that adaptive epigenetic changes could appear in germline cells and that such changes could survive the epigenetic resetting that appears to occur during meiosis.

The foundation for Jablonka and Lamb's argument is that a variety of chromosomal marking mechanisms underlie epigenetic variation. This is perhaps the most interesting part of the book, as a number of exciting results have emerged over the last ten or so years that nicely tie in with classical observations. For example, it has been known for some time that certain organisms treat chromosomes differently depending on which sex contributed them. Classic examples are inactivation of paternally-derived X-chromosomes in marsupials and the loss of all paternally derived chromosomes in certain invertebrates. Recent experiments have very convincingly shown that this behavior, called parental imprinting, can apply to particular individual genes as well as entire chromosomes. A striking example is Prader-Willi syndrome (PWS) in humans, caused by a mutation in gene on chromosome 15. There are two classes of PWS individuals, those whose father has a defective copy of PWS and those who obtain both copies of the PWS gene from their mother through a rare nondisjunction of chromosome 15. Only parental copies of this gene are expressed, so in the former event the zygote expresses a defective copy, while no expression at all occurs in the latter event. An identical DNA sequence can thus yield very different phenotypes, depending on whether it passed through the sperm or egg. There are numerous examples from transgenetic mice wherein a gene coming from one sex is inactive. Indeed, in the mouse some genes must have been received through the female and others from the male in order to form a viable zygote, presumably reflecting the inactivation of these genes in one of the sexes.

A variety of pathways for chromosomal marking are known and it is likely that several apply to any particular region of DNA. As with the lambda phage, marking can be very specific, with particular DNA segments marked by the binding of specific protein regulatory elements (e.g., transcription activators, enhancers, and suppressors) to modulate the expression of a specific gene or set of genes. Once bound, these complexes tend to assure their replication by both regulating their own expression (often suppressing expression of competing factors) and through cooperative binding that facilitates the rapid formation of complexes on newly replicated DNA. For example, the lysogenic state of phage lambda is maintained by the cI repressor gene, which binds to control regions for the lytic cycle, shutting down those regions. At the same time, cI also promotes its own transcription. When the bacteria divides, both daughter cells receive copies of the cI protein, which quickly binds to (and suppresses) the lytic control regions of both cells while also promoting its own expression, maintaining the lysogenic state.

A second pathway for chromosome marking is modification of certain DNA bases themselves, such as occurs with methylation. For many (but not all) genes, methylation levels are inversely correlated with levels of expression. DNA methylation can have specific, local effects on a few genes, so that islands of methylated (or unmethylated) genes can exist in a sea of sequence having the opposite methylation state. Conversely, its effects can be chromosome-wide, for example as seen with inactivated X-chromosomes in mammalian females. Methylation also plays a critical role in an important type of epigenetic variation that forms the basis for the bacterial immune system. Bacterial restriction enzymes, which cut at particular base sequences (restriction sites), form a primitive immune system by digesting foreign DNA such as viruses. Self versus foreign DNA is distinguished by whether the restriction sites are methylated, with foreign DNAs being unmethylated and hence cut. Active methylation systems in the bacteria ensure that newly replicated DNA has methylated restriction sites. Thus, if a virus does manage to replicate, its progeny will contain methylated bases at restriction sites and hence be immune to this particular restriction system.

A final system of chromosome marking, differences in overall chromatin structure, is not fully understood. It has been clear since the mid-1970s that subtle features of chromatin structure itself - the complex of a DNA sequence bound by histones and numerous other proteins - can influence gene expression. The exact roles played by various chromatin-associated proteins are still unclear and it is still unresolved whether chromatin changes themselves are critical to regulation or are a consequence of regulation. Nonetheless, tightly packed regions of chromatin have properties different from those of more openly packed regions and these differences are often associated with changes in gene expression. It is well known that translocations which move genes closer to regions of tightly packed heterochromatin can have rather dramatic effects on gene expression, a phenomena known as position-effect variegation. For example, the normal product ([w.sup.+]) of the white gene gives red-eyed flies and this is dominant over the white mutant (w) which gives white eyes. However, if the white locus is translocated to a heterochromatic region, [w.sup.+]w heterozygotes have eyes that are a mixture of red and white facets, as opposed to the entirely red-eyed flies that occur when the white gene is in its normal position. Molecular biologists are especially concerned with position effects in experiments involving transgenetic individuals, as the identical gene sequence inserted randomly in the genome by mobile elements (which carry the entire gene, its control regions, plus several kilobases of flanking DNA) behave very differently depending upon where they insert.

Assuming epigenetic changes through chromosome marking, Jablonka and Lamb start their argument for the possibility of Lamarckian inheritance by suggesting that environmental effects may induce stable epigenetic changes. Their argument for this is rather confusing and misleading, citing as examples some very special genetic systems: e.g., VSG-phase switching in Trypanosomes, mating-type switching in yeast, and amplification of the chorion genes in insects, to name a few. These systems all show directed DNA rearrangements that produce either a change in gene expression (VSG and mating-type switching) or amount (chorion genes). For example, the mating type of a yeast can be either type a or type alpha, but type a can only mate with alpha (and vice-versa). The yeast cell carries three DNA sequences for mating type: an active copy at the MAT locus and two silent copies at loci flanking MAT (HML and HMR which carry the sequences for alpha and a, respectively). The mating type of the yeast is entirely determined by the expressed copy at MAT, but a cell can switch its mating type by replacing (via gene conversion) the current sequence at MAT with the sequence of the other type contributed from HML or HMR.

Rather than being epigenetic changes, these examples are specific programmed changes in the DNA sequence itself and are highly evolved systems that deal with particular problems routinely faced by the organism. For example, VSG switching is an adaptation allowing trypanosomes to evade the immune systems of their hosts, mating type switching allows the descendants of a single cell to undergo sex, and chorion amplification provides a huge amount of gene product in a limited developmental window. These are not the responses to some novel environment, but rather the result of evolution to solve a particular recurring problem. What further sets these systems apart is that for most genes studied, changes in regulation do not normally involve DNA rearrangements. Thus, these examples are not directly relevant to the authors' case.

Even assuming directed mutations with an epigenetic basis do indeed occur, there remains the problem of passing these onto the next generation. Epigenetic variants face two problems in this regard. The first is the standard problem that with a multicellular organism, most of the cells involved in specific tissues cannot leave daughter cells that become part of the germline. While the authors correctly point out that many organisms do not have a strict somatic/germline distinction, they offer no solution to this problem in organisms (such as most familiar animals) that do have this strict distinction. Further, in organisms where this distinction is weak or nonexistent, there is no reason to favor epigenetic mutations over "standard" mutations due to simple changes in DNA sequence. Indeed, once an epigenetic variant appears in a germline cell, there remains the second problem of transmitting that epigenetic state intact onto the gametes.

While epigenetic changes can be stably inherited by cells undergoing mitosis, for multicellular organisms meiosis seems to reset the genome to some baseline epigenetic state from which the developmental program unfolds. For an epigenetic variant to be evolutionarily important, it must pass unchanged through this resetting. While there are striking features to this resetting (e.g., paternal imprinting), these effects usually last only a single generation, with the effect of an allele depending only on the sex of its parent, not its grandparents or more distant relatives. While the authors mention a number of examples of "lingering modifications," wherein a nongenetic effect appears to persist over several generations, there are alternative explanations for most of their examples besides epigenetic modification of DNA sequences. Most could be simple consequences of a number of cytoplasmic effects. For example, they cite Highkin's work on inbred lines of peas, where plants grown in poor environments are stunted, and also produce offspring that are stunted even when raised in a normal environment, but return to their normal height after one to two generations in the normal environment. One simple explanation for this is a reduction in the number of organelles (chloroplast and/or mitochondrial) in the offspring of the parents raised in poor environments, with organelle number taking a generation or so to recover to normal levels. Just how widespread such lingering effects are is unclear, and the authors do not present a convincing case that sequences with a long epigenetic memory (wherein effects persist for multiple generations) occur frequently enough to have a major evolutionary impact. Further, and more to the authors' point, they have not ruled out other causes but rather make the assumption that such lingering modifications could be caused by persistent chromosomal marking.

The final class of arguments offered by Jablonka and Lamb is perhaps the most interesting, dealing as it does with interactions between the genetic (DNA) and epigenetic inheritance systems. They make two main points, first that chromosome marking is very heavily influenced by repeated DNAs, so that differences in the amount of repeated DNA can change the epigenetic potential of a given region. Second, they argue that since epigenetic changes involve changes in chromosome marking, that this in turn results in different epigenetic states having potentially different mutation rates. This offers, they suggest, a route whereby an epigenetic change, by influencing the types of mutation that can occur in the DNA region involved, can be assimilated into a genetic change.

The search for functional roles for repeated DNA (which makes up over half, and usually closer to 80-90%, of all DNA in a typical organism) is the "holy grail" of adaptationist molecular biologists. This analogy with the quest for the grail is especially true in that it appears there is no such function to be found. Rather, neutral models can fully describe all the relevant features of repeated DNAs. For example, while highly repeated DNAs are the main constituent of heterochromatin, there is little evidence to suggest that these sequences are important in determining the actual chromatin structure. Rather, repetitive sequences tend to accumulate in regions of low recombination (such as heterochromatin) as suggested by Charlesworth, Langley, and others. The authors are correct in pointing out that regulatory regions for particular genes are often very short repeats and that a change in the number of such repeats can influence the regulation of that particular gene. However, there are but a few classes of such regulatory repeats in a huge universe of other classes of repeats that can arise by simple consequences of DNA replication, repair, and recombination.

As to Jablonka and Lamb's point that the epigenetic state of a DNA region can influence the mutation rate, there is at least some partial support for this. For example, molecular biologists have debated, without resolution, whether transcriptionally active genes have mutation rates different from those of inactive genes. It is certainly the case that methylated bases have a higher mutation rate than nonmethylated bases, and this can result in a mutation bias. A key issue for the authors' argument is whether different epigenetic states can persist for a sufficiently long time for any potential differences to be important. As mentioned above, there is no clear evidence that epigenetic changes due to chromosomal marking can indeed persist over several generations. Further, even if they did, there is certainly no evidence to suggest that they alter the mutation spectrum in such a way as to increase the probability that a random mutation would assimilate a region towards the epigenetic state it is currently in.

In summary, the authors have not made a convincing case that epigenetic changes, through chromosomal marking, are likely to have major evolutionary importance. Even if such a change appears in a germ cell, it will almost certainly be erased by the epigenetic resetting mechanism that appears to occur in germline cells. If their mechanism was important, one would expect to see rather rapid response to selection using completely inbred lines, as though these lines have no genetic variability (beyond a low level of new variability due to polygenic mutation), they are certainly capable of showing epigenetic variation. In reality, such lines typically show a very slow response, on the order that one would expect given known polygenic mutation rates.

In spite of these very serious reservations, I nonetheless recommend this book. Jablonka and Lamb have certainly accomplished the very desirable goal of forcing evolutionary biologists to think of inheritance as more than simply DNA sequences. However, while the DNA is certainly not naked, the effects of its epigenetic "clothes" are likely far less grandiose than the authors wish to imply.


STEELE, E. J., R. M. GORCYZNSKI, AND J. W. POLLARD. 1984. The somatic selection of acquired characters. Pp. 217-237 in J. W. Pollard, ed, Evolutionary theory: Paths into the future. John Wiley, Chichester, U.K.
COPYRIGHT 1996 Society for the Study of Evolution
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 1996 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Walsh, J. Bruce
Article Type:Book Review
Date:Oct 1, 1996
Previous Article:Energy, risk, and reptilian reproductive effort: a reply to Niewiarowski and Dunham.
Next Article:Good genes and direct selection in the evolution of mating preferences.

Terms of use | Privacy policy | Copyright © 2022 Farlex, Inc. | Feedback | For webmasters |