Epigenetic control of mobile DNA as an interface between experience and genome change

Mobile DNA in the genome is subject to RNA-targeted epigenetic control. This control regulates the activity of transposons, retrotransposons and genomic proviruses. Many different life history experiences alter the activities of mobile DNA and the expression of genetic loci regulated by nearby insertions. The same experiences induce alterations in epigenetic formatting and lead to trans-generational modifications of genome expression and stability. These observations lead to the hypothesis that epigenetic formatting directed by non-coding RNA provides a molecular interface between life history events and genome alteration.


INTRODUCTION
Understanding the functional organization of the genome and its evolutionary history are key goals of modern molecular biology. The task has become more interesting and complex as we learn more the details of cell processes involving the genome. Recent applications of high resolution technologies to genome expression in animals reveal a dynamic four-dimensional interactive control architecture incompatible with prior notions that genomes contain discrete functional segments of DNA ("genes") (Mercer and Mattick, 2013). This review will focus on the role of epigenetic regulation of viruses and mobile genetic elements as a key interface between the activities of these agents of evolutionary change and inputs from cell and organism life histories. The hypothesis developed as a result of the review is that disruption of epigenetic silencing constitutes a major target for life history activation of cellular functions for genome change. This likely occurs after genome replication, possibly by changes in small non-coding (snc) RNAs, typically on the order of 20-50 nucleotides long.

MOBILE DNA IS A MAJOR AND FUNCTIONALLY SIGNIFICANT COMPONENT OF GENOMES
One of the major surprises to come from the initial sequencing of the human genome was the high abundance of dispersed mobile repeat elements (Consortium, 2001). Today, we estimate that at least two-thirds of our genomes is composed of mobile DNA (De Koning et al., 2011). The human genome is not exceptional in its high content of mobile DNA (http://shapiro.bsd.uchicago.edu/ TableII.1.shtml).
Mobile DNA is a major source of novel coding information. One mechanism is the process known as "exonization," when splice signals are utilized in newly inserted DNA segments (http://shapiro.bsd.uchicago.edu/Origin_of_New_Protein_Domains.html). New coding sequences also form by reverse transcription of processed RNAs and genome insertion of the cDNAs, sometimes producing chimeric fusions with existing exons (http://shapiro.bsd.uchicago.edu/ Table 5B. Reports of retrogenes in plant and animal genomes.html) (Long, 2001;Betrán et al., 2002;Fu et al., 2010).
The potential functional importance of distributed mobile DNA in genomes grows rapidly as evidence accumulates for pervasive genome transcription (http://shapiro.bsd.uchicago. edu/PervasiveGenomeTranscription.html) and for the regulatory role of non-coding RNAs (ncRNAs) in genome expression, including the functional juxtaposition of distant genome regions to activate transcription (http://shapiro.bsd.uchicago. edu/NonCodingRNAinGenomeExpression.html). Mobile elements participate in this long-range genomic communication and provide the sequences of many ncRNAs (Kapusta et al., 2013).

CELLS USE RNA-TARGETED EPIGENETIC CONTROL TO INHIBIT THE ACTIVITY OF MOBILE DNA
Given the high content of mobile DNA in many genomes, an important question is: what prevents all the mobility systems from destroying genome integrity? In eukaryotic cells, a major control mechanism is sncRNA-directed epigenetic formatting into silent chromatin (Law and Jacobsen, 2010;Castel and Martienssen, 2013).
Both prokaryotes and eukaryotes have systems for capturing fragments from invading DNA molecules and placing the fragments into special loci encoding sncRNAs (Dumesic and Madhani, 2014). In prokaryotes, these loci are called CRISPRs (clustered regular interspersed palindromic repeats) (http:// shapiro.bsd.uchicago.edu/CRISPRs.html) (Marraffini and Sontheimer, 2010;Garrett et al., 2011;Bikard and Marraffini, 2013;Watanabe et al., 2013). The RNA transcripts from CRISPRs are processed into sncRNAs that target cleavage of homologous invading DNA and also inactivation of complementary mRNA (Djordjevic et al., 2012). The details of the RNA processing and interference activities are well-characterized, but the acquisition of DNA fragments is poorly understood. The process must be very rapid, because viral infection yields cells that survive the initial infection with appropriate fragments added to their CRISPR repertoire (Barrangou et al., 2007).
Virtually all eukaryotes investigated, with the notable exception of budding yeast, have mechanisms for sncRNA-directed chromatin silencing. They are based on members of the Argonaute family of sncRNA-processing proteins (http://shapiro. bsd.uchicago.edu/microRNA-directedchromatinsilencing.html). Plants and animals have independently evolved distinct mechanisms of processing the sncRNAs for the Argonaute family systems, but both groups use targeted epigenetic regulatory processes to defend against virus infection (Ding and Voinnet, 2007;Csorba et al., 2009) and prevent genome instability ( Table 1). Like prokaryotes, Drosophila has specific genomic loci where it acquires fragments of invading DNA to encode the targeting sncRNAs (Brennecke et al., 2007(Brennecke et al., , 2008Handler et al., 2013).

LIFE HISTORY EVENTS DESTABILIZE GENOMES AND ACTIVATE MOBILE DNA
Anyone who has studied real-time genome changes quantitatively knows that mutation frequencies depend upon the treatment of the experimental organism prior to measurement. A wide variety of life history events influence the natural genetic engineering (NGE) functions that generate mutations, especially mobile elements ( Table 2; Shapiro, 2011). In some cases, the genome instabilities are large scale and last multiple cell or organismal generations.
Many observations demonstrate responses of the circuits controlling NGE functions to biological and abiotic inputs. It is particularly significant that many such responses occur following exceptional cell interactions with viruses or with other cells, either by infection or by hybridization ( Table 2). As we might expect, the introduction of alien DNA or chromatin into a cell often has disruptive effects on genome homeostasis (Shapiro, 2014).  Korzeniewski et al., 2011;Akagi et al., 2014 2013). The observed epigenetic responses include alterations to cytosine methylation in DNA (Chinnusamy and Zhu, 2009b), histone modifications in nucleosomes, and sncRNAs (Ruiz-Ferrer and Voinnet, 2009;Ng et al., 2012) as well as transgenerational inheritance of complex novel phenotypes (Zucchi et al., 2012), frequently induced by stress . The phenomenon of hybrid vigor, or heterosis, increasingly is viewed as an alteration in sncRNA-targeted epigenetic formatting stimulated by the encounter of two distinct genome control regimes (Groszmann et al., 2011;Miller et al., 2012;Shivaprasad et al., 2012). Many of the studies demonstrating induced epigenetic modifications also document accompanying genome instabilities and emphasize their evolutionary potential (Madlung and Wendel, 2013). It is noteworthy that many of the same stimuli are involved in both genomic and epigenomic responses in plants (Hegarty et al., 2013) and animals (Arkhipova and Rodriguez, 2013). The common stimuli include infection and symbiosis (Hamon and Cossart, 2008;Bierne et al., 2012;Takahashi, 2014), hybridization and changes in ploidy.

DIRECT INTERACTIONS BETWEEN NGE ACTIVITIES AND EPIGENETIC REGULATORY FUNCTIONS
In addition to disruption of sncRNA-targeted inhibition, there is limited but growing evidence that NGE functions acting on DNA molecules interact directly with epigenetic control factors. There is convincing evidence of the connection between NGE and the epigenome in DNA damage repair and retroviral or retrotransposon insertions into chromosomes.
Published reports indicate that H2AX incorporation into chromatin suppresses conversion of single-strand nicks to DS breaks (Franco et al., 2006) and affects the processing of the ends of broken DNA molecules (Helmink et al., 2011). H2AX operates in phosphorylated form (Rogakou et al., 1998;Kinner et al., 2008).
Nucleosome disassembly is probably necessary for certain repair processes (Linger and Tyler, 2007;Amouroux et al., 2010;Gospodinov and Herceg, 2013), and histone modifications affect damage-induced checkpoint signaling (Chen and Tyler, 2008). Once repair is complete, nucleosome modifications are reversed, and H2AX∼P is eliminated from chromatin (Svetlova et al., 2010). So-called "bystander" cells, which are not subjected to DNA damage but are in the same culture as irradiated cells, also display H2AX phosphorylation (Sokolov et al., 2007;Dickey et al., 2009aDickey et al., , 2011. A key feature of genome repair is that H2AX-marked damaged DNA mobilizes to subnuclear "repair centers" where homologous recombination and NHEJ proteins also localize (Lisby and Rothstein, 2005;Plate et al., 2008;Bekker-Jensen and Mailand, 2010). A role for chromatin in mobilization of damaged DNA has been proposed (Seeber et al., 2013), but multiple sources of evidence are lacking.

RETROVIRAL AND RETROTRANSPOSON INTEGRASES
A more extensive case for NGE-chromatin interactions comes from analysis of retroviral and retrotransposon insertion specificities (Zhang and Mager, 2012). Each type of retrovirus displays a characteristic insertion specificity for its provirus (Lewinski et al., 2006). A number of targeting mechanisms involve epigenetic formatting molecules.
HIV and other lentiviral targeted integration into actively transcribed regions of the genome is associated with transcriptionassociated histone modifications, including H3 acetylation, H4 acetylation, and H3 K4 methylation, but is disfavored in regions rich in transcription-inhibiting modifications, which include H3K27me3 and DNA CpG methylation . The specificity results from integrase tethering by the LEDGF/p75 chromatin-binding growth factor (Vanegas et al., 2005;Llano et al., 2006;Ciuffi, 2008;Meehan and Poeschla, 2010;Zheng et al., 2010;Christ and Debyser, 2013). Replacing the LEDGF/p75 domain that interacts with expressed chromatin by the CBX1 domain, which binds histones H3K9me2 or H3K9me3 found in pericentric heterochromatin, targets HIV insertions to silent chromatin regions (Gijsbers et al., 2010).
It is probably not coincidental that the most widely distributed group of retrotransposons among all eukaryotic clades are the "chromoviruses," which are so named because they have chromodomains in their integrase proteins (Gorinsek et al., 2004;Kordis, 2005;Novikov et al., 2012;Weber et al., 2013). A chromodomain has been reported to target fungal chromovirus MAGGY insertions to heterochromatin marked by H3K9me2/me3 (Gao et al., 2008). An integrase chromodomain also participates in activator protein-targeted insertion of fission yeast retrotransposon Tf1 upstream of RNA polymerase II transcription start sites (Hizi and Levin, 2005;Chatterjee et al., 2009).

DNA TRANSPOSONS
In contrast with many retrotransposons that interact with nucleosomes, the DNA transposon Hermes inserts preferentially in budding yeast into nucleosome-free regions of the genome (Gangadharan et al., 2010). The widely used P element DNA transposons in Drosophila show targeting (called "P element homing") by incorporating binding sites for various regulatory factors, including chromatin insulators (Bender and Hudson, 2000;Fujioka et al., 2009) and Polycomb group response elements (Kassis, 2002;Cheng et al., 2012).

EPIGENETIC REFORMATTING AFTER DNA REPLICATION AND ncRNAs AS POTENTIAL AGENTS FOR TRANSMITTING EXPERIENCE TO THE GENOME
While the evidence is increasingly abundant for effects of different life history events on epigenetic regulation in general, and on genome homeostasis in particular, it is far from clear how those effects occur (Lim and Brunet, 2013). We know very little about the connections between cell sensors and epigenetic (re)formatting complexes Narlikar et al., 2013). Deciphering those connections is currently an important research goal.
DNA replication provides a key decision point for maintaining or changing chromatin configurations (Poot et al., 2005;Liu and Gong, 2011;Mermoud et al., 2011). The replication apparatus must disassemble chromatin for polymerization and then reassemble chromatin once replication is complete. Replication takes place only in dividing cells, and transgenerational inheritance of epigenetic states must involve the proliferating cells that give rise to gametes. Transfer of outside information from somatic tissues to the germline has been reported in mammals (Sharma, 2013;Skinner et al., 2013). And epigenetic windows of susceptibility to environmental insults have been suggested during sperm development (Soubry et al., 2014). Since there is no segregated germ line in plants and eukaryotic microbes, the same cells that experience environmental inputs can also be the progenitors of gametes.
One frequently overlooked feature of post-replication reestablishment of epigenetic formatting is where in the nucleus it might occur. Replication takes place in specialized "replication factories" (Vago et al., 2009;Guillou et al., 2010). Does chromatin reestablishment occur in the same location or does it involve migration of newly replicated DNA segments to distinct subnuclear "chromatin factories," like the ones that exist in the nucleolus for heterochromatin formation on rRNA-encoding DNA (Guetg and Santoro, 2012)? If so, such post-replication relocalization would be guided by the nucleoskeleton and lncRNAs (Mercer and Mattick, 2013; and might present an attractive target for stress response and sensory input signaling (Weiner et al., 2012).
It is notable that changes to ncRNAs are frequently cited with regard to the impact of life history events on the genome (Sunkar et al., 2007;Khraiwesh et al., 2012;Lelandais-Briere et al., 2012;Nakaminami et al., 2012;Amaral et al., 2013). In the plant literature, there is documentation of numerous ncRNA changes in response to particular biotic and abiotic stress regimes ( Table 4).
A number of observations about resistance to biotic and abiotic stresses are consistent with a key role for ncRNA changes in life history responses. Several viruses encode siRNA suppressors to overcome host defenses Omarov and Scholthof, 2012;Guo and Lu, 2013). Transgenic constructs encoding constitutive miRNA expression can lead to salt and drought tolerance in creeping bentgrass , to immunity against blast fungus in rice , and in Arabidopsis to greater salt and alkalinity sensitivity . Acquired aphid resistance in Arabidopsis involves sncRNA changes (Kettles et al., 2013), and most acquired stress resistances in plants display transgenerational epigenetic inheritance (Holeski et al., 2012;Luna and Ton, 2012;Slaughter et al., 2012).

SPECULATIVE CONCLUSIONS ABOUT AN EPIGENETIC INTERFACE BETWEEN EXPERIENCE AND GENOME CHANGE
Mobile DNA and other NGE functions are the key agents for adaptively significant changes in genome organization and DNA sequences. The data reviewed and tabulated above establish the importance of RNA-directed chromatin formatting in the regulation and operation of mobile elements, viruses and DNA repair

Salt
Multiple plants Ding et al., 2009;Qin et al., 2011;Macovei and Tuteja, 2012;Carnavale Bottino et al., 2013;Li et al., 2013;Ren et al., 2013;Zhuang et al., 2014 Drought Multiple plants Barrera-Figueroa et al., 2011;Li et al., 2011a;Qin et al., 2011;Wang et al., 2011;Eldem et al., 2012;Ferreira et al., 2012;Ding et al., 2013;Gentile et al., 2013;Shuai et al., 2013 Waterlogging Maize, poplar Zhang et al., 2008b;Liu et al., 2012;Ren et al., 2012;Zhai et al., 2013 Cold stress Wheat Tang  Powdery mildew infection Wheat (Xin et al., 2011) miRNAs (Xin et al., 2010) Verticillium wilt infection Cotton, eggplant Yin et al., 2012;Yang et al., 2013 functions. In addition, there is a remarkable correlation between the life history events that activate NGE functions to destabilize genomes and those that lead to alteration of chromatin states and DNA methylation patterns. The preceding observations lead to the plausible hypothesis that epigenetic regulation serves as a key interface between organismal life history and the agents that restructure genomic DNA. This hypothesis is supported by the limited number of cases where empirical observations have established direct molecular connections between NGE functions and components of the epigenetic control system: histones, nucleosomes, and chromatin reformatting complexes.
If, as I expect, further research bolsters the epigenome-NGE correlations and connections documented above, then we need to ask: what components(s) of the epigenetic control apparatus communicate information about experience to NGE operators? We do not know the answer to this fundamental question. However, the data reported in Table 4 indicate that ncRNAs are good candidates for key intermediates in the experience-genome signal transduction process. If this is so, then ncRNAs are logical molecular targets for modulating genome change toward potentially adaptive outcomes. Let us hope that research aimed at examining this proposal deepens our understanding of how life history impacts both epigenetic and genome change operations (Tables 2-4), whether or not my speculation ultimately proves to be correct.

ACKNOWLEDGMENTS
The author is grateful to the editors for the invitation to contribute to this special issue and for the opportunity to comment on the relationship between life history and genome change.