Standing guard: Perinuclear localization of an RNA-dependent RNA polymerase
Sexual reproduction has both potential benefits and potential hazards for the fitness of a genome. The genetic shuffling that accompanies meiosis is recognized as a way of testing genetic variation against environmental changes to select for evolutionary fitness. However, sexual reproduction can also assist the spread of infectious genetic parasites by intermingling two genomes. Meiosis-specific defense mechanisms have arisen to combat this hazard, and this type of defense is strikingly evident in the filamentous fungus Neurospora crassa. A recent study by Shiu et al. (1) in this issue of PNAS sheds new light on a meiotic process that targets transcripts from unpaired regions of chromosomes for destruction during meiosis.
The maintenance of a relatively stable genome is paramount to the survival of a species. Transposable genetic elements (TEs) can have immediate and drastic impacts on the structure of a genome through insertional mutagenesis and defective repair of double-stranded breaks. Given the potential danger of these elements, extensive invasion of genomes seems to have been tolerated surprisingly well in most species. To wit, 45% of the sequence of the human genome is comprised of TEs or their degenerated remnants. The situation in plants can be even more extreme, with TEs providing as much as 80% of the genomic content in some species (2, 3).
The presence of TEs in all eukaryotic genomes indicates an ancient origin and continuous maintenance during evolution. Their ubiquity and extensive contribution to genome structure have led many to propose that genome evolution and organism complexity have been driven by both the elements themselves and attempts by host genomes to tame or repress them (2, 4). Indeed, many of the mechanisms that regulate transcriptional programs in eukaryotic development also target and repress TE proliferation. These include DNA methylation, repressive chromatin assembly, and RNA-mediated interference or silencing (RNAi). These mechanisms have also been observed to be interdependent in a number of different organisms, suggesting that they have coevolved along with their original and common targets: TEs (5).
Although it has been shown that sexual reproduction is beneficial to TE spreading, and even theorized to have arisen at the urging of TEs (6), it is becoming clear that the meiotic landscape is a major battleground in genome defense. The introduction of two genomes into a common cytoplasm during sex provides an opportunity for TEs from one nucleus to infect the other: hence, the benefit to their spreading. However, meiosis also provides an opportunity for the two genomes to be compared and scanned for invaders to be targeted for repression or destruction before transmission to the next generation. This process is strikingly evident in N. crassa, in which sexual reproduction is accompanied by a premeiotic process termed RIP (for repeat-induced point mutation). RIP identifies duplicated genomic regions and targets them for extensive C-to-T mutation, usually followed by establishment and maintenance of DNA methylation and repressive chromatin assembly. Such a mutagenic process bodes equally poorly for both TEs and endogenous gene duplications and has resulted in a genome not only scrubbed clean of intact TEs but also largely devoid of paralogous gene pairs—which illustrates the potential evolutionary cost of a highly robust genome defense (7). It is not understood how duplicated regions are targeted by RIP; however, a pairing-dependent mechanism has been suggested, because only even numbers of duplicated genes are subjected to RIP (8).
Homologous chromosome alignment and synapsis during meiosis can be viewed as a chance for the sexually introduced genomes to perform a linear, gene-by-gene comparison along the length of each chromosome. Regions of differential insertions of TEs, for example, might be identified as localized misalignments, or bulges. Neurospora seems to be particularly sensitive to such misalignments, because small deletions or insertions (or extensive RIP-mediated mutation of a gene on one homolog) can target repression of the copy on the other homolog after entry into meiotic prophase (reviewed in ref. 9). This process, originally termed meiotic silencing of unpaired DNA or simply meiotic silencing, was first described by the Metzenberg laboratory. Deletions of genes required for ascospore development, the products of Neurospora meiosis, exhibited a dominant sterility phenotype. This dominance was due to the unpaired status of the unaltered gene as it passed through meiosis matched to its deleted homolog, which triggered its repression. Furthermore, silencing of all other genes with homology to the unpaired segment was observed, independent of the pairing status of the homologous targets (10, 11).
Mutational screens for defects in meiotic silencing have been performed, using the strategy of suppressing the dominant sterility of unpaired essential meiotic loci. The first mutation isolated, sad-1 (suppressor of ascus dominance), was in a gene encoding a putative RNA-dependent RNA polymerase (RdRP) (12). Cellular RdRPs are required for RNA-mediated repressive mechanisms, such as RNAi, in a number of organisms (5). Similar screens have further identified other conserved RNAi components as effectors of meiotic silencing, verifying the central role of RNAi in this process (13).
Another RdRP, QDE-1, is essential for repeat-induced posttranscriptional gene silencing (PTGS) in the vegetative phases of Neurospora (14). In contrast, the expression and activity of SAD-1 is limited to sexual stages and is required for normal meiotic progression—homozygous sad-1 strains are sterile (11). Similarly, an RdRP in the nematode Caenorhabditis elegans, EGO-1, is also only expressed in meiotic tissue, is essential for germ-line RNAi and fertility, and is required for meiotic silencing (15). Meiotic silencing has also been observed in mammals, where unpaired regions of chromosomal translocations accumulate a variety of chromatin modifications and factors involved with transcriptional repression (16, 17). In both C. elegans and mammals, endogenous targets of meiotic silencing are the unpaired or poorly paired sex chromosomes during male meiotic prophase (16, 18). The repression of the X chromosome during male meiosis by meiotic silencing mechanisms likely creates a difficult environment for male germ-line-specific genes on this chromosome, and indeed there is substantial evidence for the depletion of such genes in a number of organisms (19, 20). This depletion is consistent with genome defense mechanisms and their targets playing important roles in the shaping of genomes.
The new study by Shiu et al. (1) identifies another mutation that suppresses meiotic silencing, named sad-2. Expression of the sad-2 gene, like that of sad-1, is limited to the sexual phase. Sequence analysis of sad-2 yields no clues to its function. Functional fluorescence-tagged proteins were expressed to investigate its subcellular distribution. GFP:SAD-2 showed a striking perinuclear distribution in sexual phases, from just before karyogamy until diplotene stage in meiosis. Coexpression of GFP:SAD-2 with RFP:SAD-1 revealed extensive colocalization of these proteins at the external nuclear periphery. Furthermore, SAD-1 localization depended on SAD-2, because GFP:SAD-1 distribution was scattered in the cytoplasm in the sad-2 mutant strain. In contrast, SAD-2 distribution was relatively unaffected in the sad-1 mutant strain, indicating that SAD-2 normally resides at the nuclear periphery where it recruits SAD-1 RdRP. These results strongly suggest that SAD-1 must be localized to the nuclear periphery for its proper function in meiotic silencing, because this process is defective in the sad-2 mutant.
Why is perinuclear localization of SAD-1 required? One possibility is that SAD-1 performs a “quality” check on RNAs as they exit the nuclear pores (Fig. 1). The role of RdRPs in silencing mechanisms is not clear, but they are assumed to be important for the generation and/or amplification of double-stranded RNAs (dsRNAs). RdRPs are proposed to somehow recognize “aberrant” RNAs (aRNAs) and convert them to dsRNAs that are substrates for the dsRNA-specific Dicer endonuclease that creates the short, dsRNA effectors molecules termed siRNAs. The QDE-1 RdRP can convert ssRNA to dsRNA with or without primers and can create either complete dsRNA products or generate short dsRNA segments along the template (21). The siRNAs generated by Dicer could anneal to target sequences and be used as primers by an RdRP to generate more target-specific dsRNAs to amplify and maintain the response, as has been proposed for tandemly repeated TE silencing (22). Normal mRNA processing is exquisitely controlled, and complete assembly of a number of messenger ribonucleoprotein complex (mRNP) components is required both for nuclear export and to prevent destruction by nonsense-mediated decay pathways (23). An aRNA generated by inappropriate initiation—which in meiotic prophase may be prevented by some aspect of homolog synapsis—could become an available substrate for SAD-1 by virtue of an incomplete complement of mRNP components as it exits the nuclear pore. The dsRNA product would in turn be cleaved by Dicer to produce siRNAs capable of directing the degradation machinery (RISC) to RNAs with sequence homology: these RISC targets could now include normal mRNAs (Fig. 1).
Model for meiotic silencing in N. crassa. Aberrant RNAs generated by inappropriate transcription of unpaired DNA during meiosis are recognized as substrates by the SAD-1 RdRP anchored near nuclear pores by SAD-2. The dsRNA synthesized by SAD-1 is cleaved by Dicer to generate siRNAs. These siRNAs are used as guides by the RISC complex to degrade RNAs with sequence homology to the siRNA.
Interestingly, the perinuclear distribution of the SAD-2/SAD-1 “complex” is reminiscent of other perinuclear structures that are limited to cells involved in sexual reproduction (germ cells) in C. elegans. These structures, called “P granules,” are protein/RNA complexes that localize to the cytoplasmic face of nuclear pores during meiosis. P granules have been proposed to function in mRNA transport, translation control, and/or as carriers of maternal factors during assembly and asymmetric distribution of “germ plasm” during embryogenesis (24, 25). P granule distribution in adult meiotic cells is abnormal in ego-1 RdRP mutant animals, which may be an indirect effect, but is interesting in light of the study by Shiu et al. (26). Meiotic cells in simpler organisms such as Neurospora are presumably the evolutionary precursors to the germ cell lineage in metozoans. As the only cells that contribute to subsequent generations, germ cells are ground zero in the arms race between TEs and their metazoan hosts. Perhaps this arms race, in addition to impacting the evolution of genomes, has also contributed to the evolution of unique regulatory features of the germ line.
Footnotes
- *E-mail: bkelly{at}emory.edu
-
Author contributions: W.G.K. wrote the paper.
-
Conflict of interest statement: No conflicts declared.
-
See companion article on page 2243.
- © 2006 by The National Academy of Sciences of the USA
