RNA polymerase IV defines epigenetic variation in maize.

Paramutation is a bizarre but likely widespread phenomenon that violates Mendel’s Law of Segregation. In this process, one allele changes the activity state of its partner allele on the homologous chromosome, usually from an active state to a silent state. This new state is inherited by the next generation and beyond. What’s more, the newlymodified allele has the ability to change the activity state of naı̈ve alleles. In this process, the characteristic of a gene is “remembered” by later generations, even if the particular allele that initiated this change is no longer present. Paramutations can be transmitted meiotically as well as mitotically. Therefore, siblings can have the same genetic sequence but very different phenotypes. Paramutation was first discovered in maize (Zea mays; Brink, 1956) and was later observed in other plants aswell as in animals. A classic example of paramutation occurs with the Pl1-Rhoades haplotypes (combinations of alleles and regulatory sequences that are genetically linked) of the purple plant1 (pl1) locus in maize, which encodes an anthocyanin pigment regulator. When combined in a heterozygote, a Pl1-Rhoades haplotype in the strongly expressed Pl-Rh (purple) state becomes repressed by a homologous haplotype that exists in a weakly expressed Pl’ (colorless) state and acquires the meiotically heritable Pl’ state during development. Thus, this colorless state is meiotically inherited, or conditioned, through the generations. This process requires the cellular machinery that generates 24-nucleotide RNAs called small interfering RNAs (siRNAs), which helps maintain the repressed Pl’ state, as well as the repressed states of other maize haplotypes subject to paramutation. Indeed, siRNAs maintain the majority of methylation patterns (silencing) of repetitive genomic sequences, which comprise over 85% of the maize genome. This is not the entire story, however. While siRNAs and paramutation are closely linked, genetic studies have shown that germline transmission of siRNAs is not sufficient for facilitating paramutation in the next generation (Erhard and Hollick, 2011). Because siRNAs are generated from repetitive sequences by alternative RNA polymerase complexes, which include RNA polymerase IV (Pol IV), Erhard et al. (pages 808–819) wanted to find out what would happen if Pl1-Rhoades was transmitted through several generations of mutants lacking Pol IV function. Would the meiotically inherited silencing of this haplotype be abolished? Indeed, when Pl1-Rhoadeswas transmitted from Pol IV mutants such as rpd1 (RNA polymerase D1; which lacks the largest subunit of Pol IV), Pl1-Rhoades acquired and retained an expanded expression domain; the purple pigment was increasingly expressed over the course of several generations (see figure). Even when Pol IV function was restored, the conditioned expression pattern of Pl1-Rhoades continued. Thus, Pol IV is responsible for defining tissue-specific regulation at this specific haplotype. The authors also demonstrate that Pol IV functions by co-opting transposon-derived sequences as regulatory elements. Specifically, they found that conditioned pl1 kernel expression is correlated with the presence of a proximal CACTA-like transposon fragment. The loss of Pol IV led to the increased expression of pl1 haplotypes containing this repetitive fragment. Therefore, the absence of persistent Pol IV action at these repetitive sequences influences the gradual increase in pigmentation observed after several generations of conditioning in the rpd1 mutant background. This is an excellent, concrete example of how changes to the heterochromatic component of the genome, rather than changes to the DNA sequence, coincide with heritable changes in gene regulation. Elucidating this process is a fundamental challenge of epigenetics, of which paramutation is an extreme example.