Mouse Models of Epigenetic Inheritance: Classification, Mechanisms, and Experimental Strategies

C H A P T E R

15 Mouse Models of Epigenetic Inheritance: Classification, Mechanisms, and Experimental Strategies Shaoshuai Mao*, Yongqin Li*, Bo Liu*, Tian Chi*,** *ShanghaiTech University, Shanghai, China; **Yale University Medical School, New Haven, CT, United States

O U T L I N E Two Types of Epigenetic Inheritance (EI): Mitotic Versus Generational

231

EI Signals: Cis Versus Trans

232

Mice as Model Mammals

233

Mouse EI Models: A Brief History

233

Conventional Mouse EI Models Spontaneous Random Epigenetic Variations Experimentally Induced EI

234 234 237

TWO TYPES OF EPIGENETIC INHERITANCE (EI): MITOTIC VERSUS GENERATIONAL Transient signals, including external cues (delivered by extracellular agents, such as hormones or environmental toxicans) and cell-intrinsic “noises” (emerging stochastically from ongoing biological processes inside the cells), can trigger sustained changes in gene functions without mutating DNA. Such stable changes in the functional states of genes, called epigenetic changes, is reversible but can be durable enough to be passed onto daughter cells following mitosis, and even to the offspring of an organism if the changes occur in the germline. The propagation of epigenetic states from a somatic cell to its progeny cells, or from a gamete to the resultant Handbook of Epigenetics. http://dx.doi.org/10.1016/B978-0-12-805388-1.00015-8 Copyright © 2017 Elsevier Inc. All rights reserved.

Targeting Epigenetic Modifications to Reporter Genes: A Unique EI Model Tractable for Mechanistic Studies

238

Conclusions

240

References

241

organism, is each an example of epigenetic inheritance (EI), with former called “Mitotic EI” in this chapter and the latter, “Generational EI” [1,2] (Fig. 15.1). Generational EI includes at least two subtypes: “Intergenerational EI,” where the epigenetic state is transmitted from the individual directly experiencing environmental challenges to the immediate offspring (F1 to F2 in Fig. 15.1), and “Transgenerational EI,” defined as “the germline transmission of epigenetic information between generations in the absence of any environmental exposure” [3] (F2 to F3 in Fig. 15.1; F1 to F2 is not transgenerational because the F1 germline has been exposed). Of note, the definition of EI in this chapter echoes an early definition of epigenetics as “the study of mitotically and/or meiotically heritable changes in gene function that cannot be explained by changes in DNA sequence” [4]. However,

231

232

15.  Mouse Models of Epigenetic Inheritance: Classification, Mechanisms, and Experimental Strategies

FIGURE 15.1  A hypothetical model of EI. (A) Normal development of an embryo (F1) inside the mother (F0). The embryo consists of soma and germline, the latter conventionally considered F2. The somatic cells undergo mitosis as the embryo develops and grows into the adult. Germline development is more complex. In males, the primordial germ cells (PGCs) emerge at embryonic day 7 (E7) and become mitotically arrested “prospermatogonia” at E14. Shortly after birth, prospermatogonia become spermatogonial stem cells (SSCs), which proliferate and mature into primary spermatocytes before the latter undergo meiosis to become sperm at about 1 month after birth. In females, the occytes have already developed in the embryo, but are arrested in meiosis till fertilization. The gametes (sperm and egg) develop into the offspring via numerous steps (dashed arrow). (B) An environmental factor acting on the embryo (lightening bolt) inflicts a similar epigenetic lesion (explosion) on both the soma (e.g., brain) and the germline (top right image). The lesion in the brain survives mitosis and is transmitted to the brain of the adult F1 mouse (Mitotic EI, left panel). On the other hand, the lesion in the fetal germline survives both mitosis and meiosis to emerge in the mature gametes in the adult F1 mouse, and is subsequently passed onto the offspring via the gametes (Generational EI, right panel).

we add the term “inheritance” to “epigenetic” to emphasize the heritable nature of the EI process. This emphasis is necessary because nowadays, “epigenetic” is also loosely used to refer to any gene regulatory process that occurs at the level of chromatin, be it heritable or not [5]. We also wish to emphasize that by definition, to qualify as “epigenetically heritable,” the functional state of the gene must persist through cell division, rather than get erased during the division and then reinstated thereafter. This distinction is not only conceptually important, but also medically relevant, as seen at the human tumor suppressor MHL. A certain allele of this gene is methylated and silenced in the somatic cells in the family members, but the functional state is cleared in the germline of the parents before being reestablished in the next generation, which could dictate a different therapeutic strategy than if the silencing is epigenetically inherited [6]. Generational EI is obviously much harder to achieve than Mitotic EI, but the basic mechanisms underlying the two forms of EI are similar, as detailed here.

EI SIGNALS: CIS VERSUS TRANS In general, there are two mechanisms whereby a transient stimulus can induce heritable changes in gene function [7]. First, the stimulus causes modifications

to chromatin (histones and/or DNA), which are then recognized and perpetuated by cellular machinery capable of maintaining and replicating the modifications (Fig. 15.2A). This EI system works in cis in that the chromatin marking is a molecular signature of genes, affecting only the alleles it physically associates with. In contrast, the second EI system works in trans. Here, the transient stimulus triggers the expression of a transacting factor (such as transcription activator or RNA) capable of maintaining its own expression via, e.g., an autoregulatory loop; as the factor is freely diffusible, both the alleles will be affected (Fig. 15.2B). Mitotic and generational EI can each be mediated by cis or trans signals. For Mitotic EI, it is difficult to distinguish between the two EI signals, except in the case of monoallelic expression (e.g., genomic imprinting and X-inactivation), which by necessity is under the control of cis-signals. In contrast, the signals underlying Generational EI of a phenotype are readily distinguishable by out-crossing inbred epigenetically modified mice and their progeny to genetically identical control mice. If the phenotype in the offspring is inherited via cis-signals, its penetrance will decline progressively in successive generations as the epigenetically marked allele is increasingly diluted out by the “naïve” control allele. In contrast, the penetrance can remain unchanged if the inheritance is directed by strong, durable trans-signals.

IV.  Model Organisms of Epigenetics



Mouse EI Models: A Brief History

233

FIGURE 15.2  Maintenance and propagation of transcription states via cis- or trans-acting epigenetic signals, as represented by CpG methylation and transcription activator, respectively. (A) Cis-acting mechanism. A transient stimulus (lightening bulb) induces the methylase DNMT3-catalyzed de novo symmetric CpG methylation (CH3) on both strands of the DNA at a target gene; only one allele is shown for simplicity. Immediately following DNA replication, the DNA is hemimethylated, but the methylation tag preexisiting at the parental strand quickly helps recruit the maintenance methylase DNMT1 to copy the methylation to the daughter strand (blue), thus restoring the intact epigenetic signal. A similar mechanism might be in place if a methyl group on one of the two strands is accidentally lost in resting cells (bracket). Other chromatin features, including histone modifications, nonhistone chromatin proteins, higher-order chromatin structure, and nuclear localization are also heritable but the mechanisms underlying their heritance are poorly understood. (B) Trans-acting mechanism. A transient stimulus triggers the expression of an activator from its target gene, which then binds its own promoter to sustain its expression in the absence of the initial trigger; only one allele is shown for simplicity. During cell division, the activator, probably detached from the DNA, is partitioned to daughter cells, where it reestablishes its own expression.

MICE AS MODEL MAMMALS The laboratory mouse is the premier mammalian model organism for biomedical research. Mouse and human shares 99% of the genes and most physiological and pathological features, with similarities found in all systems including the nervous, cardiovascular, endocrine, immune, musculoskeletal systems [8]. Mice are small, docile, prolific, and easy to rear in controlled environments. Hundreds of inbred lines with divergent traits that suit various experimental purposes have arisen through a combination of natural evolution and human-directed breeding. Sophisticated mouse genetic tools have been developed to manipulate the mouse genome and define its function, which include insertion of exogenous genes via transgenesis and deletion of endogenous genes via homologous recombination [9]. Genome manipulation and phenotypic characterization are greatly facilitated with the advent of Cas9-based gene editing and high-throughput sequencing methods [10,11]. A concerted global effort, under the auspices of the International Mouse Phenotype Consortium, is in progress to “discover functional insight for every gene

by generating and systematically phenotyping 20,000 knockout mouse strains” (http://www.mousephenotype.org). Furthermore, large scale high-throughput random mutagenesis in mice is feasible using transposons (Sleeping Beauty and piggyback) or the chemical mutagen N-ethyl-N-nitrosourea (ENU), which enables both forward (gene-driven) and reverse (phenotype-driven) genetic screen [12–15]. Within the near future, the functions of the majority of protein-coding genes in mice are expected to be deciphered, and the noncoding RNAs are following suit. In many aspects, mouse is the most studied and best understood mammal, and the insights obtained from mouse research has proven invaluable for understanding human health and disease.

MOUSE EI MODELS: A BRIEF HISTORY The use of mice to study EI dates back at least to the late 19th century. At that time, the Lamarckian principle that “acquired characteristics are heritable” is accepted by some people, who would cite, as the key supporting evidence, anecdotal reports claiming that

IV.  Model Organisms of Epigenetics

234

15.  Mouse Models of Epigenetic Inheritance: Classification, Mechanisms, and Experimental Strategies

the phenotypes caused by body injury or mutilation are sometimes transmitted to the offspring. To refute the claim, August Weismann did a famous experiment in which he cut off mouse tails over successive generations and determined its effects on the offspring. Specifically, “901 young were produced by five generations of artificially mutilated parents, and yet there was not a single example of a rudimentary tail or of any other abnormality in this organ” [16]. Furthermore, as Weismann pointed out, the ceremonial mutilation of certain parts of human body from times immemorial (such as circumcision) has “not in a single instance led to the malformation or reduction in the part” [16]. Collectively these data demonstrate that the effect of mutilation is not heritable. Unfortunately, this conclusion was overly generalized to cover all acquired traits, thus putting a final end to Lamarckism. However, over the past few decades, there has been a renewed interest in Lamarckism, as evidence accumulates that ancestral environmental exposures can affect the progeny via the germline, and more strikingly, in some cases, the phenotypes developed in the offspring more or less recapitulate that directly induced by the environmental factors in the ancestors [1,2,17–24]. Although Generational EI has been observed in numerous species [23], murine models have contributed most extensively to our knowledge in this area. Mice have also contributed greatly to our knowledge about Mitotic EI, especially heritable gene silencing. Indeed, several classic examples of heritable gene silencing in mammals, namely X-inactivation, parental imprinting, and metastable epialleles (see section "Conventional Mouse EI Models" ), were all discovered in mice (between 1960s and 1990s) [25] Finally, since 1990s, after gene targeting in mice became feasible, many genes involved in epigenetic processes, including histone modifications, chromatin remodeling, and noncoding RNA metabolism, have been knocked out and their functions subsequently revealed, which is indispensable for defining the genetic basis of EI. In summary, mouse has historically served as a key model for exploring all three aspects of EI in mammals: Generational EI, Mitotic EI and the genetic basis of epigenetic control. We will outline the mouse models of Mitotic and Generational EI in the upcoming section.

CONVENTIONAL MOUSE EI MODELS For the most part, EI phenomena in mice might be classified roughly into three broad catagories. The first involves standard developmental processes, such as lineage differentiation, parental imprinting, and X-inactivation, which are regulated by elaborate developmental signals and executed by sophisticated epigenetic machinery, with well-defined biological functions. The

second type of EI phenomena are in conflict with the first. Here, the cells exposed to identical developmental signals or isogenic mice raised under identical conditions paradoxically show significant phenotypic variations. Such variations emerge randomly during normal development as a result of the noise inherent in cellular biological processes, and their physiological significance remain largely unclear. Finally, in the third type of EI phenomena, mice are experimentally challenged with environmental factors capable of fine-tuning or disrupting normal development, leading to long-lasting consequences, with obvious biological and clinical implications. The first type of EI phenomena has been extensively characterized and reviewed [26–29] (see also Chapters 18, 24, and 25), and so we will discuss only the second and third types (i.e., spontaneous random variation and experimentally induced variation), which involve noncanonical developmental processes and are poorly understood.

Spontaneous Random Epigenetic Variations We will classify the mouse models of such variations based on whether the variations are mitotically or generationally heritable. Mitotic EI Models In these models, epigenetic variations can occur at the cellular level within a mouse (variegated expression), or at the organismal level among individual mice (intangible variance). VARIEGATED EXPRESSION: DIFFERENTIAL GENE EXPRESSION IN THE CELLS OF THE SAME TYPE WITHIN A MOUSE

This is often observed in transgenic lines, where a transgene is silenced in a subset of cells of the same type [30,31]. Endogenous genes in genetically modified mice can also display variegated expression. For example, deleting a Cd8 enhancer or mutating the Cd4 silencer turns uniform CD8 and CD4 expression into variegated one [32,33]. Variegated expression at endogenous genes is clinically relevant, as exemplified in facioscapulohumeral dystrophy (FSHD), a genetic disease caused by contraction of the D4Z4 macrosatellite repeat array on chromosome 4 [34]. This contraction leads to ectopic, stochastic derepression of the transcription factor DUX4 in a small fraction of skeletal muscle cells, which causes apoptosis as well as other pathological changes in the muscle. Variegated expression has been most extensively studied at “metastable epialleles,” the mammalian alleles displaying labile epigenetic states [35,36]. Metastable epialleles are associated with retrotransposons that are young in evolutionary terms and so might

IV.  Model Organisms of Epigenetics



Conventional Mouse EI Models

235

FIGURE 15.3  Epigenetic phenomena at Avy. (A) Locus structure. Agouti signals hair follicular melanocytes to make yellow pigment instead of black. In WT mice, Agouti transcription is controlled by a promoter in exon 2 of the agouti gene (black arrow). This promoter is active only at a specific stage of hair growth, and the resultant transient AGOUTI expression causes only a subapical yellow band on each black hair. At Avy, a contraoriented intracisternal A particle (IAP) retrotransposon (broad red arrow) is inserted into the Agouti locus. A cryptic promoter in IAP drives ectopic agouti transcription(red arrow) and hence ectopic yellow coloration, but the promoter is subject to variable degrees of silencing, resulting in coat color variations. (B) Coat color spectrum in isogenic Avy mice of the same age and sex. The colors are linked to the CpG methylation status of the cryptic IAP promoter, with yellow being hypomethylated and brown fully methylated. The promoter activity can also differ among individual cells within the same mouse, producing epigenetically mosaic mouse with a mixture of yellow and brown hair (mottled). Of note, the yellow color is associated with obesity due to pleotropic effects of Agouti. (C) Generational EI of coat color. Avy females (circles) are bred with mice lacking functional agouti gene (not shown) to produce offspring. The result indicates that the coat colors are partially heritable. Furthermore, the effect lasts at least another generation (not shown). Interestingly, the father’s colors are not heritable (not shown). (D) Lasting consequences of maternal diet. Avy mice are exposed to excessive methyl donors (delivered through maternal diet) during development and lactation. The transient exposure leads to permanent shift of coat colors, which is associated with IAP CpG hypermethylation. Source: Part B, Reproduced from Jirtle, RL. The Agouti mouse: a biosensor for environmental epigenomics studies investigating the developmental origins of health and disease. Epigenomics 2014; 6: 447–450 [38]. Part C, Adapted from Rosenfeld C, Tollesfbol T, editors. Transgenerational epigenetic inheritance: evidence and debates. Elsevier; 2014: 123–145; Morgan H.D., Sutherland H.G., Martin D.I., Whitelaw E. Epigenetic inheritance at the agouti locus in the mouse. Nat Genet 1999; 23: 314–318 [60,62]. Part D, Adapted from Barouki R., Gluckman P.D., Grandjean P., Hanson M., Heindel J.J. Developmental origins of non-communicable disease: implications for research and public health. Environ Health 2012; 11:42; Waterland R.A., Jirtle R.L. Transposable elements: targets for early nutritional effects on epigenetic gene regulation. Mol Cell Biol 2003; 23:5293–5300 [70,72].

still possess some ability to escape silencing [25,37], as revealed at the prototypical metastable epiallele Agouti viable yellow (Avy) [38]. Avy carries such a retrotransposon inserted upstream of the Agouti transcription start site (Fig. 15.3A). A cryptic promoter at the retrotransposon causes aberrant Agouti transcription in the hair follicles, resulting in the yellow coat color. However, this promoter is silenced stochastically (associated with CpG methylation) in variable numbers of cells in the hair follicle, which is established before gastrulation and persists into adulthood (and even transmitted to the offspring via the eggs; see further). If the promoter is silenced in all the cells, the coat color becomes brown, but if only patches of the cells undergo silencing, the mice are epigenetically mosaic, displaying mottled colors resulting from a mixture of brown and yellow cells, which has become the poster boy of variegated expression because of its macroscopic visibility (Fig. 15.3B).

Interestingly, the degrees of silencing vary dramatically even among genetically identical littermates (Fig. 15.3B). Such differential gene expression among isogenic individuals is termed “variable expressivity” [36], a form of “intangible variance” (see following). INTANGIBLE VARIANCE: PHENOTYPIC VARIATION IN ISOGENIC LITTERMATES RAISED UNDER THE SAME CONDITION

For inbred mice raised under the standard conditions, many quantitative traits (such as body weight, size, behavior, and stress response) conform to a normal distribution (bell-shaped curve) despite tightly controlled environment and genetic background [25]. Thus, beside genotype and environment, a “third component” [39], named “intangible variance” [40], is also a key determinant of traits; in fact, intangible variance contributes to ∼80% of the variance in body

IV.  Model Organisms of Epigenetics

236

15.  Mouse Models of Epigenetic Inheritance: Classification, Mechanisms, and Experimental Strategies

weight [25,39]. Intangible variance is established by the 8-cell stage during embryogenesis, and then persists for life [39]. Multiple other cases of intangible variance have been reported, including regression of a random subset of embryos in utero [41], incomplete penetrance of phenotypes caused by mutations [42], and the aforementioned variable expressivity at Avy [38]. Variable expressivity has also been reported in transgenic lines. Indeed, shortly after the publication of the first transgenic line in 1980, it was already realized that transgene expression can vary not only among different founders due to random integration of the transgenes (i.e., position-effects), but also among different offsprings derived from the same founders, which cannot be explained by position effects [31]. The molecular basis of variable expressivity, as in the case of variegated expression, is stochastic, heritable gene silencing [25,35,36], and such silencing might also contribute to other forms of intangible variance. RELATIONSHIP BETWEEN VARIEGATED EXPRESSION AND VARIABLE EXPRESSIVITY

To our knowledge, metastable epialleles and transgenes that display variable expressivity always show variegated expressions. The reverse is true for metastable epialleles, but not for transgenes, as transgenic lines often display variegated expression without concomitant variable expressivity [30]. Furthermore, for endogenous genes, mutations at cis-regulatory elements can cause variegated expression without concomitant variable expressivity [32,33]. Thus, variegated expression seems a basic form of variation, whereas variable expressivity requires additional conditions. It is unclear why the two forms of variations are linked at metastable epialleles but not the other genes. This dichotomy is perhaps based on differences in the mechanisms of silencing. Metastable epiallele silencing occurs at the retrotransposons, and so reflect genome defense against invading parasites [43]. In contrast, transgene silencing results from adjacent heterochromatin (position effects) [44,45], repeats within transgene concatemers (repeat-induced gene silencing) [46–48], or prokaryotic vector sequences associated with the transgenes [49]. Note that of the three scenarios causing transgene silencing, the last one is reminiscent of that at metastable epialleles, as both involve foreign DNA. It would be interesting to determine whether the transgenes that display both variegated expression and variable expressivity are indeed silenced due to vector sequences. BIOLOGICAL SIGNIFICANCE OF PHENOTYPIC NOISE

Variegated expression and intangible variation are phenotypic noise rooted in stochasticity in gene expression [25]. Gene expression is intrinsically stochastic,

which arises from random fluctuations in transcription and translation due ultimately to the fact that biological processes are fundamentally driven by random collisions between macromolecules [50,51]. Noise is generally considered harmful, as it can cause death of cell or embryo and increase disease susceptibility [50]. Indeed, in normal mice, noise is actively and constantly buffered, as mutations have been identified that lead to its exacerbation. For example, mutations at the cis-acting elements turn the uniform CD4/CD8 expression into variegated one [32,33], and depletion of chromatin modifiers increases quantitative trait variations [52]. However, regulated noise and the resultant phenotypic diversification can also be beneficial by e.g., establishing heterogeneity in an initially homogenous cell population, which seems prerequisite for lineage differentiation [51,53]. Thus, noise may be biologically important [51,53,54]. Studies of variegated expression and intangible variation have illuminated, and should continue to illuminate, the mechanisms of phenotypic noise, with a wide range of implications. Generational EI Models In Generational EI, the epigenetic states established in the gametes inside the parents are (partially) transmitted to the offspring. Gametes are highly differentiated cells, and their epigenetic states must be reset shortly following fertilization (at the preimplantation stage) to confer totipotency on the early fetal cells, a prerequisite for their subsequent differentiation [55]. However, the resetting is incomplete, as demonstrated by the fact that CpG methylation at all imprinted and some nonimprinted genes are (partially) resistant to the reprogramming [56–59]. Thus, physiological mechanisms exist that enable genes to escape reprogramming, which makes Generational EI possible. Generational EI has been clearly observed at two metastable epialleles (Avy and AxinFu) [25,35,60] and four transgenes [61]. At Avy, the epigenotypes of the mothers shift the spectrum of epigenotypes of the pups: yellow mothers (with active Avy) produce 57% yellow, 43% mottled but no brown pups while brown mothers (with inactive Avy) produce 20% brown but only 40% yellow pups [62] (Fig. 15.3C). The four transgenic lines known to display naturally occurring Generational EI, E36 [63–65], 239B [66], TKZ751 [67,68], and Mta#7 [69], all carry transgenes that are expressed in the founders but heritably silenced in their offspring. The heritable silencing is established either in the germline of the founders, and/or in the germline of their children expressing genetic modifiers introduced by crossing the founders with breeders on different genetic backgrounds. The mechanisms of induction and inheritance of silencing at these transgenes are ill-defined, although CpG methylation is found associated with silencing [61].

IV.  Model Organisms of Epigenetics



Conventional Mouse EI Models

Experimentally Induced EI

237

Traditional toxicology and teratology teach us that environmental factorssuch as toxin and virus are harmful, and embryos are particularly vulnerable, as adverse gestational exposures can cause grotesque birth defects and even death of the fetus. However, over the past few decades, it has become clear that environmental factors can also act covertly, causing subtle yet lasting epigenetic defects that predispose the individual to later-onset diseases, which in humans can occur years and even decades after the exposures. Furthermore, the environmental factors can even impact the offspring for multiple generations via the germline [1,2,17–24]. This mode of action of environmental exposures may underlie the modern pandemic of many noncommunicable diseases, such as obesity, type 2 diabetes, and cardiovascular diseases [70]. Numerous mouse models have been established over the past few decades to study the long-term effects of environmental exposures. These models are highly diverse in terms of the environmental agents used, the timing of exposures, the phenotypes produced and the patterns of inheritance of phenotypes [17,19,24,60]. It is noteworthy that one of the earliest models, which is still popular today, reveals the ability of maternal dietary supplements (methyl donor) to shift the coat colors of the Avy pups [71,72] (Fig. 15.3D). Here we will focus on some crucial issues regarding the mechanisms underlying the induction and inheritance of epigenetic states, which are not usually discussed in the literature. For the sake of discussion, consider the “ideal” EI model depicted in Fig. 15.1. Here, the environmental insult induces similar epigenetic lesions in (some) somatic lineages and germline, the latter transmitted to the offspring for two generations. So far, no single animal model is known to completely conform to this hypothetical scenario, in part because testing this model would require the analysis of the epigenetic states of both the germline and the somatic cells for three generations (F1– F3), which has not been performed. However, incomprehensive analyses in diverse models suggests that some EI models do seem to fit at least portions of the “ideal” picture described in Fig. 15.1, and various models collectively would embody the entire picture, even though the evidence is largely indirect, as outlined below.

animals (F1 in Fig. 15.1), which reflect the direct effects of environmental factors on somatic cells, can be (partly) recapitulated in the F2 mice. Thus, in these models, the F1 and F2 somatic cells, and by inference, the F1 germline, presumably share some epigenetic lesions underlying the phenotypes. Indeed, prenatal viral immune activation (by injection of a viral mimic on Embryonic day 9 or E9) induces, in the ensuing adult males, behavioral abnormalities in three areas (sociability, cue fear expression, and sensorimotor gating), the first two abnormalities recapitulated in the F2 (and F3) mice [73]. Consistent with this, the expression of 2217 and 4015 genes in the amygdalar complex is altered in the F1 and F2 males, with a remarkable overlap of 1132 genes (although oddly the expression of some of these 1132 genes are changed in the opposite directions). Behavioral defects induced by postnatal stress are also (partially) heritable [74–76]. Similar to behavioral defects, some metabolic defects can be passed onto the offspring. For example, feeding young male mice with high fat diet (from 5 weeks of age for 10 weeks) induces obesity, which is recapitulated in the children (and to less extents grandchildren) [77], and a recent study using in vitro fertilization (IVF) demonstrates that such heritable effects are mediated exclusively by the gametes from obese parents [78]. The metabolic effects of poor nutrition are also partially heritable: maternal caloric restriction during gestation causes three defects in the pups (low birth weight, obesity and glucose intolerance), the first two transmitted to their offspring via the paternal line [78,79]. These data beg the question: how could environmental insults paradoxically cause similar epigenetic changes in somatic cells and the germline? To address this, it is necessary to identify primary epigenetic lesions in the somatic cells and germline. This is challenging because environmental factors are pleotropic, affecting thousands of genes most of which are presumably secondary or even tertiary targets. Further complicating the analysis, endogenous genes are often large in size and regulated by complex and poorly defined mechanisms. A mouse model where epigenetic lesions can be selectively targeted to a simple reporter gene in both somatic cells and germline would help confront these challenges (see section "Targeting Epigenetic Modifications to Reporter Genes: A Unique EI Model Tractable for Mechanistic Studies").

Induction of Epigenetic Lesion: Soma Versus Germ Perhaps the most striking aspect of the “ideal” scenario is that an environmental insult could inflict comparable epigenetic lesion on the somatic cells and germline. This is striking partly because of the dramatic differences in the ways the epigenetic states are established and maintained in the somatic cells versus germline (especially the male germline). However, many studies demonstrate that the phenotypes in directly exposed

Propagation of Epigenetic Lesion Across Generations: Mode of Inheritance and Determinant of Durability The epigenetic lesions in our hypothetic EI model can be transmitted by cis- or trans- mechanisms. Both mechanisms have been implicated in Generational EI models, but significant knowledge gaps and caveats exist. CpG methylation is perhaps the first assay one would do to analyze the epigenetic states, and indeed,

IV.  Model Organisms of Epigenetics

238

15.  Mouse Models of Epigenetic Inheritance: Classification, Mechanisms, and Experimental Strategies

where examined, aberrant methylation is often linked to the phenotypes. However, to our knowledge, in no experimentally induced Generational EI models, has CpG methylation been clearly shown to be the primary, generationally heritable epigenetic lesion (i.e., the cis-signal), and in some cases, the methylation indeed turns out to be a secondary lesion. For example, early postnatal life trauma (via maternal separation) alters DNA methylation in several candidate genes in the sperm of the separated males, and comparable changes are present in the brain of the offspring [74]. However, the aberrant methylation is not the cis-signal because the penetrance of the phenotype does not decline in the outcross experiments [74], indicating that a transmechanism is at work. Indeed, the same group subsequently found that the trauma also alters the sperm microRNA (miRNAs) in the exposed males, and injecting total RNA from these sperm into normal zygotes recapitulates the effects of trauma in the ensuing adult [80], suggesting the aberrant methylation is secondary to abnormal RNAs. In contrast to the uncertainty surrounding the cissignals, RNAs can clearly serve as the trans-signal for intergenerational EI from directly exposed individuals to the immediate offspring (F1 to F2 in Fig. 15.1). Specifically, environmental exposures such as stress and diet can alter RNA contents in the sperm, and injecting relevant RNAs from such sperm into normal fertilized eggs can reproduce the effects of environmental exposures; the RNAs injected include total RNA as aforementioned [80], a set of nine miRNAs [81] and tRNA-derived small RNAs [82,83]. These miRNAs and tRNA fragments are able to degrade stored maternal mRNAs and to regulate endogenous retroelements in the early zygotes, respectively, which presumably alter the developmental trajectory of the embryos, leading ultimately to adult phenotypes [81,82] Sperm RNAs delivered to the eggs are minute in quantities, and mammals do not have RNA-dependent RNA polymerase to replicate RNA as do lower organisms. Thus, as the eggs divide, the paternal RNAs rapidly disappear. Indeed, while early life stress alters miRNAs in the sperm of the exposed males (F1 in Fig. 15.1), the defect is absent from F2 sperm [80]. Paradoxically, in this model, the F2 sperm can pass on the behavior phenotype to F3, suggesting that the aberrant RNAs in F1 sperm have relayed the information to unknown carriers in the F2 sperm [80]. Sperm RNAs might use another strategy to affect F3, as illustrated by the following study [84]. Specifically, injecting a single miRNA (miR-1) into fertilized eggs suffices to cause cardiac hypertrophy in 90% of the ensuing adults (F1 in Fig. 15.1), which is heritable for at least 3 generations (from F2 to F4) with no decline in penetrance. Amazingly, miR-1 is elevated in both F2 and F3 sperm (F1 sperm somehow not tested), suggesting that the miR-1 injected into the eggs, which should

rapidly vanish, is regenerated in the F1, F2, and F3 sperm presumably via an autoregulatory loop (Fig. 15.2B) [84]. Of note, although it is clear that RNAs can be sufficient to replace the environmental factors in triggering the phenotypes, their necessity has yet to be shown; this would require depleting the RNA in sperm in the exposed mice. In addition to the nature of the signals directing the inheritance of epigenetic lesions, another mystery surrounding these lesions concerns their durability. In our hypothetical scenario, the epigenetic lesion in the F1 germline can be faithfully transmitted to F2 and F3. This is certainly not generally true. For example, prenatal exposure to environmental toxicans [on E8-20 during primordial germ cell (PGC) reprograming] induces multiple alterations in transcription and methylation profiles in the ensuing prospermatogonia, but none of these changes are inherited by the prospermatogonia of the immediate offspring [85]. The failure to inherit some epigenetic lesions is also consistent with the observation that often, only a subset of the environmentally induced phenotypes is generationally heritable. So what determine the durability of epigenetic lesions? Our lack of knowledge here is glaring. This issue is hard to address in part due to the pleotropic nature of environmental factors and complexity of endogenous genes. Further complicating the problem, the pleotropic epigenetic alterations in conventional EI models can trigger secondary DNA mutations [86]. These limitations again calls for mouse models allowing for selective induction of epigenetic lesions at simple target genes.

TARGETING EPIGENETIC MODIFICATIONS TO REPORTER GENES: A UNIQUE EI MODEL TRACTABLE FOR MECHANISTIC STUDIES To overcome the aforementioned limitations in conventional EI models, we have established mouse models where heritable chromatin states can be pharmacologically induced at simple reporter genes in both somatic cells and the germline [87]. Our models, quite consistent with the “ideal” model depicted in Fig. 15.1, greatly facilitates dissection of various parameters potentially influencing the establishment and propagation of epigenetic lesions, including timing of exposure, nature of epigenetic modifications, and genomic location of the target genes. These models are based on the well-defined tetracycline (tet)-sensitive gene regulatory system. Initially, we used mice ubiquitously expressing the reverse tetregulated transcription activator (rtTA) and carrying a transgene bearing its cognate DNA binding sites (tet-O) upstream of the human CMV minimal promoter and GFP (Fig. 15.4A); the transgene is inserted as a singlecopy gene into the Col1a1 locus via flippase-catalyzed

IV.  Model Organisms of Epigenetics



Targeting Epigenetic Modifications to Reporter Genes: A Unique EI Model Tractable for Mechanistic Studies

239

FIGURE 15.4  Epigenetic phenomenon at the tetO-CMV-GFP reporter gene. (A) The experimental system consists of Dox-activated transcription factor rtTA, which is ubiquitously expressed from a transgene inserted into the Rosa26 locus, and a simple reporter gene bearing rtTA binding site (TetO) upstream of the CMV minimal promoter and GFP, which is inserted into the Col1a1 locus. Depicted is the scenario in adult mice, where Dox reversibly induces GFP. (B) Fetal Dox exposure leads to widespread GFP silencing in the ensuing adults, which is transgenerationally heritable via the female germline. Prenatally exposed females (F1) are mated with naive, nontransgenic males to generate F2. F2 females are in turn mated with control males to generate F3. Adult mice are challenged with Dox before imaging. Of note, brain is not included in the photographs because Dox can not induce GFP in adult brain even in the control mice, which is due to blood-brain barrier in adults. However, the transgene in the brain from prenatally exposed mice is highly methylated, suggesting it is also silenced [87]. (C) EI pattern at the tetO-CMV-GFP reporter. Fetal Dox exposure induces widespread epigenetic silencing, but for simplicity, only the brain and the germline are shown affected. The silencing can be transmitted to F5 (unpublished). (D) Effects of rtTA on the tetO-CMV-GFP transgene in ES cells expressing rtTA. Cells are cultured in the presence of Dox for various days before analysis. The top panel displays micrographs of ES cells following 2 (left) and 15 (right) days of Dox exposure. The bottom panel is flow cytometrical quantification of GFP expression, with Dox stimulation times and percentage of GFP-expressing cells indicated at left and right, respectively. (E) Fetal Dox exposure facilitates CD4 induction in the ensuing adult mice. (Top) CD4 transcription is controlled by enhancer (E), promoter (P) and silencer (S). tetO (red box) is inserted into the Cd4 locus so that Cd4 expression is subject to regulation by Dox. The numbers denote the distance from the transcription start site (arrow). (Bottom) Flow cytometrical quantification of CD4 expression in CD8 lymphocytes in mice expressing rtTA. CD4 is normally repressed in CD8 cells. Dox administration (via drinking water) induces CD4, but slowly and inefficiently, with only 46% CD8 cells expressing CD4 following even prolonged (17 days) induction (left panel). Prenatal Dox exposure (E0-20) dramatically increases both the kinetics and the extent of CD4 induction: the proportion of CD4-expressing CD8 cells already reaches 53% within 2 days of induction, and by Day 17, the majority of CD8 cells (85%) are expressing CD4 (right panel). Source: Parts B, D, and E, Reproduced from Wan M, et al. Inducible mouse models illuminate parameters influencing epigenetic inheritance. Development 2013; 140: 843–852 [87].

site-specific recombination [88]. rtTA binds and activates target genes only in the presence of tet or its derivative doxycycline (Dox). The mice were created in the Jaenisch lab, who demonstrated that Dox administration (via drinking water) activates GFP expression in adult mice [88]. Our goal was to determine whether transient Dox preexposure in adults can facilitate the second round of Dox induction later. We observed no such epigenetic memory. Unexpectedly, when the mice were preexposed to Dox during fetal development, the gene became completely refractory to Dox challenge in the ensuing adults, and furthermore, the refractory state was maternally

transmitted to ∼30% of offspring for at least two generations (Figs. 15.4B–C). The silencing is associated with DNA hypermethylation and loss of an activating histone mark (H3K4me2) at the CMV promoter in the somatic cells in the directly exposed mothers (F1) and their offspring (F2 and F3), raising the possibility that similar epigenetic lesions have been induced in the F1 somatic cells and eggs and transmitted to F2 eggs. Furthermore, the silenced transgene can not affect the “naïve” transgene, demonstrating that the inheritance is not mediated by a trans signal. Thus, fetal Dox exposure causes transgenerationally heritable chromatin marking at this simple,

IV.  Model Organisms of Epigenetics

240

15.  Mouse Models of Epigenetic Inheritance: Classification, Mechanisms, and Experimental Strategies

single-copy reporter gene. To our knowledge, this is the first time a cis-acting mechanism is clearly shown to underlie the inheritance in experimentally induced Generational EI models. How rtTA paradoxically triggers epigenetic silencing is unclear, but the phenomenon was seen only in fetus and ES cells, where the silencing was preceded by GFP induction, suggesting it is an indirect effect of rtTA activation that requires tissue-specific repressors [87] (Fig. 15.4D). We then used this model to define the timing of establishment of the silenced states. We exposed the mice to Dox at various stages of fetal development. Dox exposure for the first 4 days was critical but insufficient for producing transgenerationally heritable silencing, but exposure for 10 days was partially sufficient. Since epigenetic reprogramming in PGCs begins around E8 in mice, the epigenetic silencing must have been largely established before the onset of this reprogramming, and once established, it must be partially resistant to the reprogramming. We have thus identified E0-10 as an essential and partially sufficient time window to induce transgenerationally heritable epigenetic lesion, which has never been reported for other EI models. To dissect other parameters influencing Generational EI, we inserted into the Col1a1 locus a very different Dox-responsive transgene, which contains the CD4 regulatory elements (promoter/enhancer/silencer) instead of the CMV minimal promoter. Despite the genetic differences, fetal Dox exposure similarly produced a transgenerationally heritable phenotype, but of the opposite nature: facilitation of GFP induction associated with activating histone modifications (H3K9a and H3K4me2). This is the first demonstration that activating chromatin marks could be transgenerationally heritable in mammals. Since both activating and repressive perturbations are heritable at the Col1a1 locus, and since the two transgenes are genetically distinct, the Col1a1 locus (rather than the nature of epigenetic modifications or the DNA sequences of the transgenes) seems the key determinant of Generational EI. In support of this, fetal Dox exposure also caused mitotically heritable silencing of a randomly integrated CMV promoter (not shown) and triggered mitotically heritable activation of a Dox-responsive endogenous Cd4 allele (Fig. 15.4E), but neither effect was intergenerationally heritable. These data also demonstrate that fetal epigenome is extremely malleable, where epigenetic perturbations, once induced at proper times during embryogenesis, can generally be transmitted to adults regardless of the nature or location of the perturbations. Consistent with this, a Dox-regulated transcription repressor (tTS) can cause mitotically stable silencing at two distinct, randomly integrated promoters if and only if the repressor acts during the first few days of fetal development [89]. Thus, mitotically heritable epigenetic perturbations can be readily established as long as the

inducers act at the critical times during embryogenesis, whereas Generational EI additionally requires special locus environment. What is special about the Col1a1 locus? Our current hypothesis is that the locus harbors insulator-like elements that protect epigenetic modifications (whether silencing or activating) from reprogramming enzymes and efforts are underway to identify the putative elements. In summary, we have established a unique approach to study EI, which has already provided insights into the mechanisms of induction and durability of cis-epigenetic signals that are hard to obtain with conventional EI models. Clearly, this approach is also applicable to transsignals.

CONCLUSIONS Ever increasing numbers of mouse models have been established to probe long-term effects of environmental factors on human health and to illuminate basic EI mechanisms, but many crucial issues remain, a few of them discussed below. First, environmental factors (mainly psychological stress and nutritional imbalance), even when acting postnatally, can apparently produce similar epigenetic changes in the soma and the germline, despite the dramatic differences between the two. Psychological stress and nutritional imbalance are not evolutionary novel, and so the organisms may have evolved mechanisms not only to confront these challenges themselves, but also to prepare the offspring for the same challenges predicted to befall them. Thus, the generational EI in such cases may be part of a physiological adaptive response (predictive adaptive response), that is hardwired in the genome [90]. It will be of great interest to decipher the mechanisms of such a response, which, to Weismann, must be “immensely complex, nay! altogether inconceivable” [16]. However, given the rapid progress in the field, it is not inconceivable that the mechanisms will be revealed in the near future, which would surely make both Weismann and Lamarck roll in their graves. Second, the parameters governing the induction and transmission of cis- or trans-epigenetic signals are poorly understood, partly due to the complexity of the endogenous genes and confounding secondary and parallel epigenetic lesions that presumably coexist with the primary epigenetic signals in conventional EI models. Mouse models with targeted epigenetic lesions would be invaluable here. Finally, little is known about the genetic basis of EI, including how genetic backgrounds affect EI, an issue of medical importance given the genetic diversity of humans. Large-scale screens using ENU have successfully uncovered multiple genetic modifiers of epigenetic

IV.  Model Organisms of Epigenetics



REFERENCES

reprograming [25,41,91–97], but this approach is laborintensive and costly. An alternative strategy would be to use mosaic analysis, which requires much fewer mice. Such screens may be feasible given the availability of lentiviral sgRNAs libraries and Cas9-expressing mice.

Glossary Epigenetic inheritance (EI) signals  Signals responsible for EI, based on chemical modifications of DNA and/or histone (cis-signal), or on diffusible molecules, such as RNAs (trans-signal). Epigenetic inheritance  Transmission of epigenetic information through cell divisions. Epigenetics  The study of mitotically and/or meiotically heritable changes in gene function that cannot be explained by changes in DNA sequence. Generational EI, intergenerational EI, transgenerational EI There seems no consensus on their definitions. We propose, based in part on literature survey, to define the three terms as “germline transmission of epigenetic information between generations,” “germline transmission of epigenetic information from an environmentally exposed organism to the immediate offspring,” and “the germline transmission of epigenetic information between generations in the absence of any environmental exposure,” respectively (see main text). Intangible variance Phenotypic variation among genetically identical individuals raised under the same condition (e.g., littermates of inbred mice). Variable expressivity is a form of intangible variance. Metastable epialleles  Epigenetically labile alleles displaying variegated expression and variable expressivity due to variable epigenetic modifications, which are stochastically established during early development. Mitotic EI  Transmission of epigenetic information from parental cells to daughter cells through mitosis. Variable expressivity  Variable expression of a gene between genetically identical individuals. Variegated expression  Variable expression of a gene in a single cell type within the same individual.

References [1] Skinner MK. Environmental epigenetic transgenerational inheritance and somatic epigenetic mitotic stability. Epigenetics 2011;6:838–42. [2] Tollefsbol T. . In: Tollesfbol T, editor. Transgenerational epigenetic inheritance: evidence and debates. Boston, MA: Elsevier; 2014. p. 1–8. [3] Skinner MK. What is an epigenetic transgenerational phenotype? F3 or F2. Reprod Toxicol 2008;25:2–6. [4] Riggs A, Porter T. Overview of epigenetic mechanisms. In: Russo VEA, Martienssen R, Riggs AD, editors. Epigenetic mechanisms of gene regulation. Cold Spring Harbor, NY: Cold Spring Harbor Laboratory Press; 1996. p. 29–45. [5] Felsenfeld G. A brief history of epigenetics. Cold Spring Harb Perspect Biol 2014;6:a018200. [6] Hitchins MP, et al. Dominantly inherited constitutional epigenetic silencing of MLH1 in a cancer-affected family is linked to a single nucleotide variant within the 5′UTR. Cancer Cell 2011;20:200–13. [7] Bonasio R, Tu S, Reinberg D. Molecular signals of epigenetic states. Science 2010;330:612–6. [8] Beckers J, Wurst W, de Angelis MH. Towards better mouse models: enhanced genotypes, systemic phenotyping and envirotype modelling. Nat Rev Genet 2009;10:371–80.

241

[9] van der Weyden L, White JK, Adams DJ, Logan DW. The mouse genetics toolkit: revealing function and mechanism. Genome Biol 2011;12:224. [10] Kaczmarczyk L, Jackson WS. Astonishing advances in mouse genetic tools for biomedical research. Swiss Med Wkly 2015;145:w14186. [11] Singh P, Schimenti JC, Bolcun-Filas E. A Mouse geneticist’s practical guide to CRISPR applications. Genetics 2015;199:1–15. [12] Carlson CM, et al. Transposon mutagenesis of the mouse germline. Genetics 2003;165:243–56. [13] Ding S, et al. Efficient transposition of the piggyBac (PB) transposon in mammalian cells and mice. Cell 2005;122:473–83. [14] Dupuy AJ, Akagi K, Largaespada DA, Copeland NG, Jenkins NA. Mammalian mutagenesis using a highly mobile somatic sleeping beauty transposon system. Nature 2005;436:221–6. [15] Kile BT, Hilton DJ. The art and design of genetic screens: mouse. Nat Rev Genet 2005;6:557–67. [16] Weismann, A. The Supposed Transmission of Mutilations. A lecture delivered at the Meeting of the Association of German Naturalists at Cologne, September 1888; 1888. [17] Bohacek J, Mansuy IM. Molecular insights into transgenerational non-genetic inheritance of acquired behaviours. Nat Rev Genet 2015;16:641–52. [18] Burggren W. Epigenetic inheritance and its role in evolutionary biology: re-evaluation and new perspectives. Biology 2016;5:24. [19] Hanson MA, Skinner MK. Developmental origins of epigenetic transgenerational inheritance. Environ Epigenet 2016;2:dvw002. [20] Heard E, Martienssen RA. Transgenerational epigenetic inheritance: myths and mechanisms. Cell 2014;157:95–109. [21] Jablonka E. Epigenetic inheritance and evolution The Lamarckian dimension. New York: Oxford University Press; 1999. [22] Jablonka E. Epigenetic inheritance and plasticity: the responsive germline. Prog Biophys Mol Biol 2013;111(2–3):99–107. [23] Jablonka E, Raz G. Transgenerational epigenetic inheritance: prevalence, mechanisms, and implications for the study of heredity and evolution. Q Rev Biol 2009;84:131–76. [24] Rando OJ, Simmons RA. I’m eating for two: parental dietary effects on offspring metabolism. Cell 2015;161:93–105. [25] Blewitt M, Whitelaw E. The use of mouse models to study epigenetics. Cold Spring Harb Perspect Biol 2013;5:a017939. [26] Lee JT, Bartolomei MS. X-Inactivation, imprinting, and long noncoding RNAs in health and disease. Cell 2013;152:1308–23. [27] Barlow DP, Bartolomei MS. Genomic imprinting in mammals. Cold Spring Harb Perspect Biol 2014;6:a018382. [28] Boland MJ, Nazor KL, Loring JF. Epigenetic regulation of pluripotency and differentiation. Circ Res 2014;115:311–24. [29] Brockdorff N, Turner BM. Dosage compensation in mammals. Cold Spring Harb Perspect Biol 2015;7:a019406. [30] Martin DI, Whitelaw E. The vagaries of variegating transgenes. Bioessays 1996;18:919–23. [31] Palmiter RD, Brinster RL. Germ-line transformation of mice. Annu Rev Genet 1986;20:465–99. [32] Garefalaki A, et al. Variegated expression of CD8 alpha resulting from in situ deletion of regulatory sequences. Immunity 2002;16:635–47. [33] Taniuchi I, Sunshine MJ, Festenstein R, Littman DR. Evidence for distinct CD4 silencer functions at different stages of thymocyte differentiation. Mol Cell 2002;10:1083–96. [34] Tawil R, van der Maarel SM, Tapscott SJ. Facioscapulohumeral dystrophy: the path to consensus on pathophysiology. Skelet Muscle 2014;4:12. [35] Kellermayer R. . In: Tollesfbol T, editor. Transgenerational epigenetic inheritance: evidence and debates. Elsevier; 2014. p. 69–73. [36] Rakyan VK, Blewitt ME, Druker R, Preis JI, Whitelaw E. Metastable epialleles in mammals. Trends Genet 2002;18:348–51.

IV.  Model Organisms of Epigenetics

242

15.  Mouse Models of Epigenetic Inheritance: Classification, Mechanisms, and Experimental Strategies

[37] Whitelaw E, Martin DI. Retrotransposons as epigenetic mediators of phenotypic variation in mammals. Nat Genet 2001;27:361–5. [38] Jirtle RL. The Agouti mouse: a biosensor for environmental epigenomics studies investigating the developmental origins of health and disease. Epigenomics 2014;6:447–50. [39] Gärtner K. A third component causing random variability beside environment and genotype. A reason for the limited success of a 30 year long effort to standardize laboratory animals? Int J Epidemiol 2012;41:335–41. [40] Falconer D. Introduction to quantitative genetics. 3rd ed. New York, NY: Wiley; 1989. [41] Blewitt ME, et al. An N-ethyl-N-nitrosourea screen for genes involved in variegation in the mouse. Proc Natl Acad Sci USA 2005;102:7629–34. [42] Biben C, et al. Cardiac septal and valvular dysmorphogenesis in mice heterozygous for mutations in the homeobox gene Nkx2-5. Circ Res 2000;87:888–95. [43] Crichton JH, Dunican DS, MacLennan M, Meehan RR, Adams IR. Defending the genome from the enemy within: mechanisms of retrotransposon suppression in the mouse germline. Cell Mol Life Sci 2014;71:1581–605. [44] Liu C. Strategies for designing transgenic DNA constructs. Methods Mol Biol 2013;1027:183–201. [45] Wilson C, Bellen HJ, Gehring WJ. Position effects on eukaryotic gene expression. Annu Rev Cell Biol 1990;6:679–714. [46] Garrick D, Fiering S, Martin DI, Whitelaw E. Repeat-induced gene silencing in mammals. Nat Genet 1998;18:56–9. [47] Henikoff S. Conspiracy of silence among repeated transgenes. Bioessays 1998;20:532–5. [48] McBurney M. Evidence for repeat-induced gene silencing in cultured mammalian cells: inactivation of tandem repeats of transfected genes. Exp Cell Res 2002;274:1–8. [49] Kjer-Nielsen L, Holmberg K, Perera JD, McCluskey J. Impaired expression of chimaeric major histocompatibility complex transgenes associated with plasmid sequences. Transgenic Res 1992;1:182–7. [50] Kaern M, Elston TC, Blake WJ, Collins JJ. Stochasticity in gene expression: from theories to phenotypes. Nat Rev Genet 2005;6:451–64. [51] Pujadas E, Feinberg AP. Regulated noise in the epigenetic landscape of development and disease. Cell 2012;148:1123–31. [52] Whitelaw NC, et al. Reduced levels of two modifiers of epigenetic gene silencing, Dnmt3a and Trim28, cause increased phenotypic noise. Genome Biol 2010;11:R111. [53] Whitelaw NC, Chong S, Whitelaw E. Tuning in to noise: epigenetics and intangible variation. Dev Cell 2010;19:649–50. [54] Eldar A, Elowitz MB. Functional roles for noise in genetic circuits. Nature 2010;467:167–73. [55] Feng S, Jacobsen SE, Reik W. Epigenetic reprogramming in plant and animal development. Science 2010;330:622–7. [56] Borgel J, et al. Targets and dynamics of promoter DNA methylation during early mouse development. Nature 2010;42:1093–100. [57] Guibert S, Forné T, Weber M. Global profiling of DNA methylation erasure in mouse primordial germ cells. Genome Res 2012;22:633–41. [58] Hackett JA, et al. Germline DNA demethylation dynamics and imprint erasure through 5-hydroxymethylcytosine. Science 2013;339:448–52. [59] Smallwood SA, et al. Dynamic CpG island methylation landscape in oocytes and preimplantation embryos. Nature 2011;43:811–4. [60] Rosenfeld C, Tollesfbol T, editors. Transgenerational epigenetic inheritance: evidence and debates. Boston, MA: Elsevier; 2014. p. 123–45. [61] Kaundal R, Yang Y, Nottoli T, Chi T. . In: Tollesfbol T, editor. Transgenerational epigenetic inheritance: evidence and debates. Boston, MA: Elsevier; 2014. p. 75–83.

[62] Morgan HD, Sutherland HG, Martin DI, Whitelaw E. Epigenetic inheritance at the agouti locus in the mouse. Nat Genet 1999;23:314–8. [63] Hadchouel M, Farza H, Simon D, Tiollais P, Pourcel C. Maternal inhibition of hepatitis B surface antigen gene expression in transgenic mice correlates with de novo methylation. Nature 1987;329:454–6. [64] Pourcel C, Tiollais P, Farza H. Transcription of the S gene in transgenic mice is associated with hypomethylation at specific sites and with DNase I sensitivity. J Virol 1990;64:931–5. [65] Schweizer J, Valenza-Schaerly P, Goret F, Pourcel C. Control of expression and methylation of a hepatitis B virus transgene by strain-specific modifiers. DNA Cell Biol 1998;17:427–35. [66] Kearns M, Preis J, McDonald M, Morris C, Whitelaw E. Complex patterns of inheritance of an imprinted murine transgene suggest incomplete germline erasure. Nucleic Acids Res 2000;28:3301–9. [67] Allen ND, Norris ML, Surani MA. Epigenetic control of transgene expression and imprinting by genotype-specific modifiers. Cell 1990;61:853–61. [68] Pickard B, et al. Epigenetic targeting in the mouse zygote marks DNA for later methylation: a mechanism for maternal effects in development. Mech Dev 2001;103:35–47. [69] Sutherland HG, et al. Reactivation of heritably silenced gene expression in mice. Mamm Genome 2000;11:347–55. [70] Barouki R, Gluckman PD, Grandjean P, Hanson M, Heindel JJ. Developmental origins of non-communicable disease: implications for research and public health. Environ Health 2012;11:42. [71] Wolff GL, Kodell RL, Moore SR, Cooney CA. Maternal epigenetics and methyl supplements affect agouti gene expression in Avy/a mice. FASEB J 1998;12:949–57. [72] Waterland RA, Jirtle RL. Transposable elements: targets for early nutritional effects on epigenetic gene regulation. Mol Cell Biol 2003;23:5293–300. [73] Weber-Stadlbauer U, et al. Transgenerational transmission and modification of pathological traits induced by prenatal immune activation. Mol Psychiatry 2016;22(1):102–12. [74] Franklin TB, et al. Epigenetic transmission of the impact of early stress across generations. Biol Psychiatry 2010;68:408–15. [75] Rodgers AB, Morgan CP, Bronson SL, Revello S, Bale TL. Paternal stress exposure alters sperm MicroRNA content and reprograms offspring HPA stress axis regulation. J Neurosci 2013;33:9003–12. [76] Saavedra-Rodríguez L, Feig LA. Chronic Social Instability Induces Anxiety and Defective Social Interactions Across Generations. Biol Psychiatry 2013;73:44–53. [77] Fullston T, et al. Paternal obesity initiates metabolic disturbances in two generations of mice with incomplete penetrance to the F2 generation and alters the transcriptional profile of testis and sperm microRNA content. FASEB J 2013;27:4226–43. [78] Huypens P, et al. Epigenetic germline inheritance of diet-induced obesity and insulin resistance. Nat Genet 2016;48:497–9. [79] Jimenez-Chillaron JC, et al. Intergenerational Transmission of Glucose Intolerance and Obesity by In Utero Undernutrition in Mice. Diabetes 2009;58:460–8. [80] Gapp K, et al. Implication of sperm RNAs in transgenerational inheritance of the effects of early trauma in mice. Nat Neurosci 2014;17:667–9. [81] Rodgers AB, Morgan CP, Leu NA, Bale TL. Transgenerational epigenetic programming via sperm microRNA recapitulates effects of paternal stress. Proc Natl Acad Sci USA 2015;112:13699– 704. [82] Sharma U, et al. Biogenesis and function of tRNA fragments during sperm maturation and fertilization in mammals. Science 2016;351:391–6. [83] Chen Q, et al. Sperm tsRNAs contribute to intergenerational inheritance of an acquired metabolic disorder. Science 2015;351(6271):397–400.

IV.  Model Organisms of Epigenetics



243

REFERENCES

[84] Wagner KD, et al. RNA induction and inheritance of epigenetic cardiac hypertrophy in the mouse. Dev Cell 2008;14:962–9. [85] Iqbal K, et al. Deleterious effects of endocrine disruptors are corrected in the mammalian germline by epigenome reprogramming. Genome Biol 2015;16:59. [86] Guerrero-Bosagna C, Skinner MK. Environmentally induced epigenetic transgenerational inheritance of phenotype and disease. Mol Cell Endocrinol 2011;354(1–2):3–8. [87] Wan M, et al. Inducible mouse models illuminate parameters influencing epigenetic inheritance. Development 2013;140:843–52. [88] Beard C, Hochedlinger K, Plath K, Wutz A, Jaenisch R. Efficient method to generate single-copy transgenic mice by site-specific integration in embryonic stem cells. Genesis 2006;44:23–8. [89] Wiznerowicz M, et al. The Kruppel-associated box repressor domain can trigger de novo promoter methylation during mouse early embryogenesis. J Biol Chem 2007;282:34535–41. [90] Hanson MA, Gluckman PD. Early developmental conditioning of later health and disease: physiology or pathophysiology? Physiol Rev 2014;94:1027–76. [91] Bultman SJ, Gebuhr TC, Magnuson T. A Brg1 mutation that uncouples ATPase activity from chromatin remodeling reveals an

[92]

[93] [94]

[95]

[96]

[97]

essential role for SWI/SNF-related complexes in beta-globin expression and erythroid development. Genes Dev 2005;19:2849–61. Percec I, Plenge RM, Nadeau JH, Bartolomei MS, Willard HF. Autosomal dominant mutations affecting X inactivation choice in the mouse. Science 2002;296:1136–9. Percec I, et al. An N-ethyl-N-nitrosourea mutagenesis screen for epigenetic mutations in the mouse. Genetics 2003;164:1481–94. Schumacher A, Faust C, Magnuson T. Positional cloning of a global regulator of anterior-posterior patterning in mice. Nature 1996;384:648. Wilson L, et al. Random mutagenesis of proximal mouse chromosome 5 uncovers predominantly embryonic lethal mutations. Genome Res 2005;15:1095–105. Jun JE, et al. Identifying the MAGUK protein Carma-1 as a central regulator of humoral immune responses and atopy by genome-wide mouse mutagenesis. Immunity 2003;18:751–62. Carpinelli MR, et al. Suppressor screen in Mpl-/- mice: c-Myb mutation causes supraphysiological production of platelets in the absence of thrombopoietin signaling. Proc Natl Acad Sci USA 2004;101:6553–8.

IV.  Model Organisms of Epigenetics