THERE WERE NO JACKASSES IN THE UNITED STATES WHEN IT WAS founded, yet soon they were everywhere. Where did they all come from?
It goes back all the way to George Washington. Ever curious, Washington kept himself abreast of what was going on elsewhere in the world, particularly as it related to farming. At some point he heard about the amazing exploits of creatures called mules and sought to bring some over from Europe for his own scrutiny. At the time, Spain had a near monopoly on mules, a legacy of the Moors. Actually, there was no monopoly on mules as such; Spain was willing to share mules with the world. The monopoly was on the means of producing mules, a tricky procedure, since mules are not created by conventional means, that is, by other mules. Mules are, rather, the spawn of the unnatural couplings of horses and donkeys. By “unnatural” I mean these things don’t just happen when donkeys and horses share a pasture; it takes some coaxing, not least finding two to tango. To increase the odds of success, the human interveners generally seek male donkeys and female horses, because the libido of a male donkey is higher than that of a typical male horse and female horses are less particular than female donkeys—though in their defense, mares are typically blindfolded.
The mules that are the result of this lack of discrimination cannot perpetuate themselves because they are sterile. Their sterility does not deter male mules from attempting to mate; they are as ardent as their donkey fathers but can never be fathers themselves. Mules must be generated anew from further indiscriminate acts between donkeys and horses. So what Washington wanted wasn’t a shipload of mules but one of those libidinous male donkeys, so he could make mules of his own. Domestic donkeys are descended from members of the horse family known as asses. For reasons obscured by the veil of time, male asses are called jacks and female asses are called jennies, while all other members of the horse family, including zebras, are called stallions and mares respectively. Hence the term jackass, its pejorative connotations deriving from the fact that donkeys of both sexes are less pliable—though by most accounts more intelligent—than horses.
Spain treated their donkeys like the Chinese did silkworms: their export was outlawed. But in 1785, Washington was, by virtue of his prestige, able to persuade King Charles III to part with one jackass named Royal Gift, on which the American mule industry was founded.1 His unnatural progeny proved invaluable in settling the country, especially as it expanded westward. Mules were particularly valued for hauling and plowing, activities for which they were favored over horses because of their superior strength and sure-footedness. Despite these virtues, mules are memorialized in this country primarily for their stubbornness and ornery disposition. According to William Faulkner, a mule will “work patiently for you for ten years for the chance to kick you once.”2 Elsewhere, however, mules have been known for their physical feats—not their behavioral drawbacks—dating back to the time they were first created over three thousand years ago in the Middle East, where there was a plentiful supply of both asses and horses.
The early mule breeders also sometimes mated horse stallions with jennies, the progeny of which are called hinnies. And from these early days, significant differences between mules and hinnies were noticed. Mules are larger and stronger than hinnies and possess larger donkey-like ears. Indeed, mules look like oversized donkeys with longer legs. Hinnies, on the other hand, are much more horselike in appearance and more tractable. (Hinnies, not mules, are deployed at Disneyland to pull carriages, for example.)
The mule-hinny divide is a 3,000-year-old puzzle, only recently solved. In the process of solving this puzzle, scientists have uncovered a novel transgenerational epigenetic effect called genomic imprinting.
It Depends on Whether It Came from Mom or Dad
The mule-hinny puzzle boils down to this: Both are half-horse and half-mule, so why should they be so different? It violates one of Mendel’s fundamental laws of inheritance. We were taught in high school biology that, aside from the Y chromosome, each parent bestows on us separate but equal genetic complements. From each parent, we get one set of chromosomes and hence genes; inheritance is sexually symmetrical. The mule was the first evidence that, in addition to the Y chromosome, there is something else asymmetrical in what is bestowed on us by our mothers and fathers. This asymmetry came to be known as the parent-of-origin effect. It was most obvious in hybrids like the mule. For example, tigons (the progeny of male tigers and female lions) and ligers (male lions and female tigers) are also quite different animals.
While most obvious in hybrids, the parent-of-origin effect has been identified through a variety of other means as well. In humans, Turner syndrome provides a case in point. Turner syndrome is a condition that results when part or all of one X chromosome is missing. Normal females, as we saw in Chapter 8, inherit one X chromosome from each parent. Turner females lack one of the X chromosomes, so their complement of sex chromosomes is denoted XO. Given the discussion about random X-chromosome inactivation in Chapter 8, you might not expect this to be a problem. Normal XX females, after all, have one functional X chromosome per cell. But recall also that the entire X chromosome is not inactivated; some genes normally escape inactivation.
Many problems of XO females can be traced to the 15 percent of the genes on the X chromosome that normally escape X inactivation. In normal XX females, both the maternal and paternal copies of these genes are expressed in all cells. In XO females, only one copy is available. This is probably why 98 percent of Turner females are spontaneously aborted. Nonetheless, Turner syndrome is present in one out of every 2,500 female live births, making it one of the more common major genetic defects.
Turner females who survive birth are subject to a number of maladies, most characteristically a failure to sexually mature. Other problems associated with the syndrome, to varying degrees, are poor growth, cardiovascular disease, osteoporosis, diabetes, and problems with spatial cognition.3 Which of these ailments a given Turner female has depends in part on whether the remaining X chromosome came from her mother or her father.4
Turner syndrome provides only limited insight into parent-of-origin effects because so much of the genome is missing. Prader-Willi syndrome (PWS) is more useful in this respect. PWS, too, is associated with a host of developmental abnormalities; these typically include obesity, poor muscle tone, undeveloped gonads, small stature, and cognitive deficits.5 There is more than one way to generate PWS, but most who suffer this malady have lost a small chunk of chromosome 15, which geneticists call a deletion.6 Within the deleted sequence are several genes and nongenic sequences (DNA that isn’t part of a gene). It is not surprising that a deletion of this sort would significantly affect development; what is surprising, though, is that this deletion results in PWS only when it is inherited from the father. If the same deletion on chromosome 15 is inherited from the mother, a completely different disorder, known as Angelman syndrome (AS), results.7 It’s as if the mother’s genes in this region have different stamps than the fathers’ genes. Both maternal and paternal stamps are required for normal development.
It is the providence of these stamps and not simply being double stamped that matters, as demonstrated by those cases of PWS in which there are no deletions. About 25 percent of PWS cases result from a different kind of molecular screwup, in which two copies of the maternal chromosome are produced, rather than the normal condition of one maternal chromosome and one paternal chromosome.8 In these cases, it is especially clear that a paternal stamp on certain genes located on chromosome 15 is required for normal development.
Some of the relevant genes in PWS/AS have been identified, but it is from studies of a third disorder, Beckwith-Wiedemann syndrome (BWS), that a particular stamped gene has been best characterized. As it happens, this gene is IGF2, which we first met in discussing the Dutch famine (Chapter 1). Recall that IGF2 is a growth factor that is especially important during fetal development.
When IGF2 has the paternal stamp, it is active; when it has the maternal stamp, it is inactive. This is the normal condition. It is noteworthy that a gene for a protein that inhibits IGF2 actions is also stamped. But the situation is reversed stamp-wise. When the IGF2 inhibitor has the maternal stamp, it is active; when it has the paternal stamp, it is inactive. This, too, is the normal condition.9 When these parent-of-origin stamps are missing, bad things can happen, one of which is BWS.
Beckwith-Wiedemann syndrome is a growth disorder resulting in an overgrowth of the fetus. BWS is also associated with several other traits, including an increased risk for a particular kind of kidney cancer, called Wilms’ tumor.10 It occurs when either IGF2 or its inhibitor is improperly stamped. But what is the nature of these stamps? And how are they established?
Parent-of-Origin Stamps Equal Genomic Imprinting
For most genes, both the version (that is, the allele) inherited from the mother and the one inherited from the father are expressed, when they are expressed at all. This typical condition is called biparental expression. For about 1 percent of our genes, however, only one of the two alleles is normally expressed. Sometimes it’s the allele inherited from the mother; sometimes it’s the allele inherited from the father. This is called uniparental expression. Uniparental gene expression occurs when either the maternal or paternal gene is more or less permanently disabled. This disabling process, formerly known as “genetic imprinting,” is now called genomic imprinting.11 Imprinting is an epigenetic process in which methylation figures prominently.
But the imprinting process is distinctive in several respects. First, there is the timing. As we saw in Chapter 7, most epigenetic alterations are removed during the process of making eggs and sperm. Imprinted genes are no exception; the epigenetic imprints are erased early in sperm and egg development. But there is a second stage of reprogramming in these reproductive cells. In this second stage, the imprinted methylation patterns are restored to the sperm and egg prior to their maturation, and hence are present during fertilization.12
The imprinted gene still must survive a second round of reprogramming, a global demethylation that occurs between fertilization and implantation.13 Imprinted genes are special in that they don’t become completely demethylated during this second round of reprogramming. Other epigenetic processes prevent this from happening. So by the time the embryo implants, imprinted genes are already epigenetically fixed in their expression pattern. This is a good thing because imprinted genes generally do most of their work early in development, long before birth.14
The reason the term genomic imprinting has come to replace “genetic imprinting” is that the imprint is neither on the gene itself, nor its control panel, nor even on a stretch of DNA adjacent to the gene. Instead, the methylation imprint can be located quite a distance away from the gene whose expression it controls, in what are called imprinting control regions (ICRs).15 In the Prader-Willi case, the ICR epigenetically regulates a number of genes on chromosome 15. Genomic imprinting shares with X inactivation this “remote control” of numerous genes.
There is another epigenetic oddity about imprinted genes: the methylation, which is the imprint, does not always block a gene’s expression; sometimes it even enhances it. So uniparental expression could arise because one allele is imprinted “on” or because the other allele is imprinted “off.” In what follows, I will simply refer to the “active allele” and the “inactive allele.” The imprinted IGF2 allele is normally active only when inherited from the father. The imprinted allele for the IGF2 inhibitor is active only when inherited from the mother.
The Role of Imprinted Genes in Development
The majority of active imprinted alleles are maternal in origin. Many of these genes are expressed in the placenta and put a brake on embryonic growth.16 Many of the paternally imprinted genes, on the other hand, seem to promote embryonic growth.17 In the rare cases in which all paternal imprints are lost, the placenta is undeveloped. Conversely, when all maternal imprints are lost, the placenta is unnaturally large. The imprinting of IGF2 and its inhibitor illustrates this contrast in microcosm. When IGF2 is improperly imprinted, such that it is expressed in both alleles rather than one, the fetus experiences the overgrowth characteristic of BWS. The overgrowth is especially pronounced if the maternally imprinted inhibitor is not present.18 Both occur when there is a duplication of the paternal portion of the chromosome and consequent loss of the maternal parts.19
Imprinting malfunctions that cause underexpression of the paternally imprinted IGF2 allele and/or overexpression of the gene for the IGF2 inhibitor result in growth retardation, as in Silver-Russell syndrome.20 So the paternally imprinted IGF2 and the maternally imprinted inhibitor work in an antagonistic fashion, and a balance between the two is required for normal development. This seems to be more generally the case as well. For normal embryonic development, a balance is required between the activities of maternally and paternally imprinted genes.21
Imprinted genes are uniquely vulnerable to molecular mishaps because of their monoallelic (one allele) expression. In most genes, which are biallelic (expression by both alleles), if something goes wrong with one allele, the other can partially compensate. There can be no such compensation in imprinted alleles. If something goes wrong, it goes wrong in a bigger way than in most genes.22 The consequences of these epigenetic mishaps are enormous, partly because they occur so early in development, but also because the messed-up imprints are more likely to be transmitted to future generations than other messed-up epigenetic processes. Imprinting has transgenerational effects.
Environmental Effects on Imprinted Genes
There is increasing interest in the effects of environmental toxins on epigenetic processes in general, and recently, genomic imprinting in particular. Here I will focus on one group of toxins called endocrine disruptors. As the name implies, endocrine disruptors disrupt physiological processes that involve hormones, typically because they mimic the hormones and bind to their receptors. Some of the most pernicious endocrine disruptors mimic the female hormone estrogen. These include polychlorinated biphenyls (PCBs) and bisphenol A, which is used in the production of plastics (such as the ubiquitous water bottles). Other estrogen-like endocrine disruptors include agricultural weed killers such as atrazine, and fungicides such as vinclozolin.
The effects of endocrine disruptors were first noticed in fish and amphibians, and are a major cause of the declines in some local populations.23 Fish and amphibians are especially susceptible for two reasons: because they live in aquatic habitats where these chemicals become concentrated, and because their sexual development is influenced more by their environment than is the sexual development of humans and other mammals.24 For example, endocrine disruptors can cause fish to change sex, resulting in all-female populations.25 They can also have a dramatically feminizing effect on amphibians resulting in male sterility.26
While less dramatic than in fish and amphibians, endocrine disruptors have been linked to a variety of ailments in humans and other mammals. The effects of endocrine disruptors on imprinted genes in mammals are especially well studied.27 Male mammals, including human males, seem particularly sensitive to developmental errors caused by the effect of endocrine disruptors on imprinted genes, as evidenced by increased rates of prostate cancer, kidney disease, and abnormal testes.28 In many cases, these problems don’t become manifest until adulthood, as in the adult-onset diseases such as the metabolic syndrome. As if this were not alarming enough, it has been recently demonstrated in rats that these defects can be transmitted to future generations.
Male rats exposed to the fungicide vinclozolin in utero have defective sperm and reduced fertility as adults. Their male offspring—with no vinclozolin exposure—also have defective sperm and low fertility, as do the males of the third and fourth generations.29 Vinclozolin exerts these transgenerational effects by altering the imprinting process during sperm development. The fungicide not only alters normal imprints but establishes new ones in parts of the genome that are not usually imprinted.30 These new imprints are transmitted through the male line for at least four generations. The new imprints not only affect fertility; they are also associated with a number of adult-onset diseases of the testis, prostate, kidney, and immune system.31
These experiments have not been replicated in humans, nor will that ever happen—what prospective mother is going to volunteer for vinclozolin exposure? But they provide compelling evidence that endocrine disruptors are not just a problem for fish and frogs.
The Hybrid Problem
We began this chapter with the mule-hinny puzzle, to which we now return. We should first note that members of the horse family have a remarkable ability to produce healthy young through hybridization. This is true not only for horses and asses, but for zebras too. You can cross a zebra with a horse and get a zorse (male zebra × female horse) or a horbra (male horse × female zebra). That is not at all the norm for mammals. Except among the most closely related species, mammalian hybrids exhibit all kinds of developmental defects and health problems, a phenomenon known as hybrid dysgenesis. Members of the horse family are not immune to hybrid dysgenesis, as is evident in the sterility of the mules and other hybrids. Traditionally hybrid dysgenesis was attributed to genetic incompatibilities. Once two species have genetically diverged sufficiently, any hybrid will suffer because when the two parental genomes are combined in the fertilized egg, they cannot be properly coordinated.
There is undoubtedly much truth to this view, but recent research suggests that it is only part of the story. Hybrid mammals also experience disruptions in the imprinting process, including a complete loss of imprinting for some genes. The loss of imprinting has been especially well demonstrated in rodents such as members of the genus Mus, which includes the common house mouse. Alleles that are normally expressed only when inherited from the mother—or from the father, as the case may be—are now expressed no matter from which parent they were derived. This can create a host of problems, beginning very early in development.32 The problem here is not so much genetic divergence as epigenetic divergence, which leads to problems in epigenetic reprogramming.
Horses and asses have also epigenetically diverged, though not to the point of causing fundamental disruption of epigenetic reprogramming and consequent loss of imprints. Stallions and jackasses bequeath slightly different imprints to their progeny, as do mares and jennies. Hence the crosses, though genetically symmetrical, are not epigenetically symmetrical. The differences between mules and hinnies quite effectively illustrate the power of this epigenetic asymmetry.
Mule Variations
Mules (and hinnies) were first created over three thousand years ago by some enterprising—if somewhat perverse—inhabitants of Mesopotamia, the first recorded example of a parent-of-origin effect. Over the years, other parent-of-origin effects were uncovered, not only in hybrids but in the transmission of numerous sorts of developmental defects, such as Prader-Willi and Turner syndromes. Yet these parent-of-origin effects remained a puzzle, even long after the advent of modern genetics. Mendel’s framework and its subsequent elaborations provided insufficient resources for understanding this phenomenon.
It is only very recently, with the advent of epigenetics, that we have at hand an explanation for mules and hinnies and other parent-of-origin effects, now known as genomic imprinting. Imprinting resembles, in some ways, the kind of epigenetic inheritance we explored in Chapter 7. An important difference is that the epigenetic mark in imprinting is not directly transmitted to the next generation as it is in the mouse agouti allele or the fwa allele in Arabidopsis. Instead, it is erased during epigenetic reprogramming, then reestablished anew. For this reason, imprinting is not considered true epigenetic inheritance, even though imprints are certainly epigenetic and they are inherited, albeit in a different manner than genes and epigenetic marks such as fwa. Whether we want to call it epigenetic inheritance or simply another kind of transgenerational epigenetic effect, imprinting clearly calls for an expansion of our notion of biological inheritance. Imprinting is a form of biological inheritance; it just follows different rules than genetic inheritance.
But genomic imprinting is first and foremost a novel form of epigenetic control of the process of development, the process whereby a fertilized egg, or zygote, became you or me. We now turn to the more common means by which this process is epigenetically regulated. For most biologists, it is in understanding development, especially very early development, that the biggest payoff for epigenetics will be.