Beyond the central dogma

The central dogma, DNA makes RNA makes protein, has long been a staple of biology textbooks. More recently, this paradigm has been extended from individual genes to whole genomes by advances in genomic technologies. For example, probing of DNA microarrays accomplishes on a large scale what was previously achieved for single genes using filter hybridization. High-throughput technology, not breakthrough biology, is becoming synonymous with genomics. The longevity of the central dogma has also meant that basic computational tools for analyzing sequence data reached maturity before whole genomic sequences became available. For example, gene finding systems were introduced over a decade ago (Fields and Soderlund, 1990), and the importance of the problem attracted many computational biologists. As a result of numerous incremental advances made already, diminishing returns may be expected for this problem. A similar situation appears to hold for database searching (Schaffer et al., 2001). Thus, the computational tools that are most widely used now may be difficult to improve upon in the postgenomic era. Technologies based on textbook biology will continue to generate opportunities in bioinformatics. However, more exciting prospects may come from new discoveries that extend or even violate the central dogma. Consider developmental biology. The central dogma says nothing about the differences between the cells in a human body, as each one has the same DNA. However, recent findings have begun to shed light on how these differences arise and are maintained, and the biochemical rules that govern these differences are only being worked out now. The emerging understanding of developmental inheritance follows a series of fundamental discoveries that have led to a realization that there is more to life than the central dogma. The central dogma was first challenged by the discovery of reverse transcription (Baltimore et al., 1970; Temin and Mizutani, 1970). Thought at the time to be peculiar to retroviruses, we now know from large-scale sequencing that our genome contains an order of magnitude more copies of sequences encoding reverse transcriptase than sequences encoding all other proteins combined (Lander et al., 2001)! Half of our genome is devoted to retroelements and their remnants, compared to only a few percent devoted to gene coding regions. Humans are not alone in having genomes dominated by retroelements. The genomes of many plants are even more infested: for example, retrotransposons occupy about 80% of the maize genome (SanMiguel et al., 1996). With so much genomic territory taken over by selfish elements, they are prime candidates for involvement in important genetic processes. One example is the propagation of silencing along the inactivated X chromosome of mammalian females: abundant LINE-1 retrotransposons were proposed to act as ‘way-stations’ (Lyon, 1998). Evidence in support of this idea was obtained by analyzing genomic sequence data (Bailey et al., 2000), an illustration of how thinking about genetic mechanisms creates opportunities in bioinformatics. Selfish elements also reveal evolutionary processes that continue to shape genomes: arguably the major scientific story of the draft human genome sequence was the history of retrotransposon evolution (Lander et al., 2001), a story missed by others who may have been so focused on the genes that they overlooked the junk (Venter et al., 2001). The success of selfish DNA elements does not mean that our genomes are entirely at their mercy. A widespread view is that genomes are protected by an immunity system (Yoder et al., 1997). Among the weapons that are thought to help protect genomes, especially in plants, are DNA methyltransferases, enzymes that mark sequences for silencing by covalent modification. Silencing of retrotransposon transcription, which must precede reverse transcription and integration, should be an effective defense against their mobilization. A major unsolved problem has been the basis for recognition of transposons and their ilk by the DNA methylation machinery. Without obvious sequence cues, it has been difficult to understand how a genome defense system protects against invaders. This question goes beyond DNA methylation: organisms such as the fruit fly, which has an almost unmethylated genome, may effectively prevent transposition by packaging retroelements in silent chromatin (van Steensel et al., 2001). Recently, a surprising solution to the problem of retroelement recognition has been proposed: RNA interference (RNAi). First elucidated in the nematode, where genes could be shut down by introduction of double-stranded RNA, this powerful gene silencing technique is now known to utilize enzymatic machinery that is common to animals and plants (Carthew, 2001). Small interfering RNAs (siRNAs) of only 22-25 bp can traverse intracellular spaces to enter cells and trigger rapid degradation of homologous RNAs. The same mechanism underlies post-transcriptional gene silencing (PTGS) in plants, where unintended post-transcriptional silencing of transgenes has been the bane of genetic engineers for over a decade. Thus, PTGS appears to be a natural mechanism for defending against RNA-based invaders. In addition, PTGS may be involved in the recognition and targeting of genomic DNA sequences: siRNA made in the cytoplasm would be targeted to the nucleus where it guides a DNA methyltransferase to covalently modify