Dissecting the Mechanisms of Posttranscriptional Gene Silencing Divide and Conquer

genes were tested. In all cases, the transcripts of homolIt is now routine to construct trangenes with defined ogous plant genes became degraded following infection promoters and terminators and express them in a variety (reviewed in Bruening, 1998). PTGS also allows plants of organisms. However, some introduced transgenes to “cure themselves” of infecting viruses even if the are not expressed as they should be. For example, virus has no homology to the host genome. In several epigenetic gene silencing inactivates expression of examples (reviewed in Carrington and Whitham, 1998) transgenes in higher plants at frequencies of up to 30% viruses infect a plant and initially replicate, but as the of independent transformants. Although many examples infection spreads, the host plant begins degrading the of transgene silencing have been described in diverse viral transcripts until no virus particles can be detected, recipient species from plants to fungi to animals (rea process called “recovery.” The “recovered” plant is viewed in Vaucheret et al., 1998), the gene regulation no longer susceptible to further infection by sequencemechanisms behind the process are poorly understood. related viruses. Not surprisingly, some plant viruses Gene silencing affects homologous sequences, for exhave evolved mechanisms to protect themselves against ample, repeats in transgenes or transgenes with sePTGS in their hosts. For example, expression of particuquence homology to endogenous genes. Two types of lar gene products of several RNA viruses has recently gene silencing mechanisms have been observed in been shown to prevent PTGS of unrelated transgenes plants, transcriptional gene silencing (TGS) and postas well as to enhance replication of other viruses (Anantranscriptional gene silencing (PTGS). The promoters of dalakshmi et al., 1998; Brigneti et al., 1998; Kasschau TGS silenced genes are inactivated so that their tranand Carrington, 1998). scription is inhibited. In contrast, genes silenced by Why are both transgenes and viruses susceptible to PTGS continue to be transcribed, but polyadenylated PTGS? Dougherty and colleagues (Lindbo et al., 1993) transcripts are nearly undetectable. Both TGS and PTGS suggested that transgenes and viruses sometimes prosilenced genes tend to be hypermethylated. TGS siduce “aberrant” RNAs distinct from normal polyadenylenced genes are methylated in their promoters, while lated transcripts, which stimulate a sequence-specific PTGS silenced genes tend to be methylated in the codRNA degradation mechanism. What constitutes an “abing region. The role of methylation in gene silencing is errant” RNA is still a matter of debate. According to still in dispute, but the prevalence of hypermethylated the “aberrant RNA model,” structurally similar aberrant sequences in silenced genes indicates its importance. transcripts derived either from viruses or transgenes can It is still not clear whether TGS and PTGS occur by stimulate the activity of RNA-directed RNA polymerases independent mechanisms, but mounting evidence sug(RdRps), plant proteins known to be activated in plants gests that TGS may have mechanistic similarities to infected with RNA viruses. Since purified RdRps can PTGS (Mette et al., 1999). The remainder of this review copy most template RNAs into small complementary will focus on PTGS. RNAs (Schiebel et al., 1998), the aberrant RNA model PTGS involves a sequence-specific RNA degradation suggests that activated RdRps recognize the aberrant process that affects all highly homologous transcripts. RNAs as substrates to be copied. If the small compleTherefore, silencing of a transgene by PTGS will also mentary RNAs produced hybridize to sequence-related silence expression of other sequence-related genes, functional transcripts, the resulting double-stranded even endogenous genes unlinked to the transgene inRNAs would be rapidly degraded by double strand– sertion, leading to rapid degradation of their transcripts. specific RNases as illustrated in Figure 1c. Although it This effect (called cosuppression) was first described has not been proven, this model has been generally in higher plants, but similar events (called quelling) that accepted because it accounts for sequence-specific involve PTGS have been described in filamentous fungi RNA degradation by the activity of known enzymes. In (Romano and Macino, 1992) and ciliates (Ruiz et al., fact, an RdRp activated upon virus infection has recently 1998). A similar phenomenon described in animals, been cloned from tomato (Schiebel et al., 1998). Homolocalled RNA interference, may function through the gous sequences were found in all plants examined and same mechanism as PTGS. Injection of double-stranded in yeast and C. elegans. It should now be possible to RNAs into cells of Caenorhabditis elegans (Fire et al., directly test the aberrant RNA model by analysis of mu1998) or Drosophila melanogaster (Kennerdell and Cartants with defects in RdRp enzymes, which should not thew, 1998) leads to inhibition of expression of homolobe able to support PTGS. In addition, a number of mugous endogenous genes. The broad range of species tants defective in PTGS have been identified in Arabiin which introduction of foreign DNA or RNA leads to dopsis and Neurospora. If the aberrant RNA model is inhibition of the expression of endogenous genes sugcorrect, we expect that some of the mutations will fall gests that gene silencing is part of a universal gene in genes with homology to the cloned RdRp or to double