Exploring and explaining epigenetic effects.

STEVEN HENIKOFF AND MARJORI A. MATZKE* steveh@muller.thcrc.org mmatzke@oeaw.ac.at HOWARD HUGIfES MEDICAL INSTrlL'TE. FRED Ht~CHfNSOX CAXCER RLSEAaCH CENTER, 1100 FAIR',IP~," AVENtSE N0.tm[, SEATrLE, WA 98109-1024, USA. *[NSq'ITUq'E OF MOLECULAR BIOLOGY, AUS'IXlAN ACADE.',IY OF SCIENCES, BILLROTH~,TI~SSE 11, A-5~20 SALZBURG, Acs'rRIA. Epigenetic phenomena in diverse or- ganisms comprise some of the most intriguing and actively investigated problems in genetics. The term 'epi- genetics' was introduced by Conrad Waddington to describe changes in gene expression during develop- ment I. Nowadays, epigenetics in the Waddington sense refers to alterations in gene expression without a change in nudeotide sequence. However, this definition is so broad that an issue of Trends in Genetics devoted to epigen- etics would read more like a modem biology textbook than a series of criti- cal reviews. A more focused descrip- tion of epigenetics refers to modifi- cations in gene expression that are brought about by heritable, but poten- tially reversible, changes in chroma- tin structure and/or DNA methylation. This issue of Trends in Genetics explores both old and new epigenetic phenomena, such as paramutation and genomic imprinting, respectively. Although no consensus has yet emerged about the source of epigen- etic effects, common threads connect- ing many of the phenomena suggest to us that host responses to mobile elements provide a unifying theme. In applied research, host reac- tions to invasive elements might be provoked by genetic transformation, causing methylation-associated gene silencing that can frustrate attempts to engineer plants and animals geneti- cally. Altered gene expression can also be the consequence of processes that do not involve nuclear epigenetic inheritance. Cosuppression in plants is a post-transcriptional silencing phe- nomenon that is due primarily to en- hanced cytoplasmic turnover of trans- gene and homologous endogenous gene RNAs (Ref. 2). Although co- suppression has not yet been de- tected in animal cells, the phenom- enon of 'quelling' in Neurospora provides a parallel (E.U. Selker, this issue). Prims represent a form of cytoplasmic inheritance involving a heritable change in state (in this case, protein conformation) 3 , but this behavior is clearly demarcated from cases of nuclear epigenetic inheri- tance (J.B. Hollick, J.E. Dorweiler and V.L. Chandler, this issue). Epigenetic effects usually involve gene silencing. In any differentiated cell, most genes are normally in- active and many of the molecular components involved in generating and maintaining this state are being discovered by studying epigenetic phenomena in yeast and Drosophila. In yeasts, where powerful genetic and biochemical approaches can be combined, the study of mating type silencing, and telomeric- and centromeric-position effects has al- lowed a thorough analysis of individ- ual constituents of multiprotein com- plexes that establish the silenced state. New epigenetic effects at several loci have indicated that these classical silencing phenomena might only represent the tip of the iceberg. The dynamics of silencing, including pro- tein modifications and changes dur- ing the cell cycle, account for much of the recent excitement in this field (J.M. Sherman and L. Pillus, this issue). In Drosophila it is possible to investigate protein components of silencing complexes in the context of well-studied developmental path- ways. There are dozens of genetically identified proteins that are involved in ,:ilencing the same target genes. These heterogeneous proteins might interact at composite binding sites and nucleate the formation of silenc- ing complexes encompassing nearby sequences. The 'stickiness' required for this to occur might also operate between homologs and even be- tween unlinked sites (V. Pirrotta, thb issue). This general model might also account for the ability of hetero- chromatin to silence reporter genes, a phenomenon known as position effect variegation (PEVE In many organisms, gene silenc- ing is associated with repeated DNA sequences. Heterochromatin in higher 'riG AUGUST 1997 VOL. 13 NO. 8 eukaryotes comprises many types of repeats, including simple sequence arrays and inactive mobile elements. However in filamentous fungi, repeti- tive sequences are almost nonexist- ent owing to the occurrence of two processes, RIP (repeat-induced point mutation) and MIP (methylation- induced premeiotically). Linked and unlinked sequence duplications can both be modified, genetically by RIP and epigenetically by both RIP and MIP. These processes clearly play genome defense roles because they only occur during a brief period of the sexual cycle and not during vegetative growth. Mobile elements are the likely natural targets (E.U. ~lker, this issue). Fungi also have epigenetic phe- nomena that resemble those seen in higher . ukatyotes. For example, meiotic tmtsvection is a normal gene regulatory process that takes place at'ross homologsS, similar to the numerous examples of transvection between mutant alleles in Drosophila. Furthermore, epigenetic states can be meiotically inherited in fission yeast 6.7 as is observed with paramutation in plants, which involves the meintically heritable weakening of one allele by the other after their interaction in the heterozygote. Such meiotic befit- abflity is a hallmark of many plant epigenetic phenomena, and is not ob- served for post-transcriptienal silenc- ing seen in cosuppression. Paramutation is a fascinating phe- nomenon because it demonstrates that some alleles and homologous unlinked loci can interact in tram, resulting in persistent changes in ex- pression after the interacting genes are inherited separately in progeny. Para- mutation thus transgresses Mendel's first law, which disallows residual influences on alleles following their segregation into different gametes. Initially believed to affect only a few plant genes, paramutation might be more common than previously imagined. Recent examples include