The RNA interference pathway can inhibit the expression of specific genes. It now seems that an essential component of the silencing process is the gene-expression machinery itself.
Molecular biologists have been amazed in recent years by the discovery of an RNA-mediated mechanism for inhibiting the expression of specific genes — the RNA interference (RNAi) pathway. The ‘RNA-induced silencing complex’ (RISC) contains small interfering RNAs (siRNAs) whose sequence of nucleotide bases can pair with that of a particular messenger RNA, targeting this mRNA for destruction before it can be translated into protein1. However, in many organisms this ‘post-transcriptional’ gene silencing is only part of the story: production of the mRNA can be shut off by a second siRNA complex called RITS (for ‘RNA-induced transcriptional silencing’).
Schramke et al. (RNA-interference-directed chromatin modification coupled to RNA polymerase II transcription)2 and Kato et al. (writing in Science)3 now show that a gene must first be transcribed if it is subsequently to be silenced. More surprisingly, this transcription must be specifically carried out by RNA polymerase II (RNApII), the enzyme responsible for making mRNAs in eukaryotic organisms.
DNA is packed into nuclei by being wrapped around histone proteins to form nucleosomes. RITS represses transcription by recruiting a histone methyltransferase to the target genes. This enzyme modifies histones so as to make the wrapped DNA inaccessible to the gene-expression machinery, creating a silenced nucleosome configuration known as heterochromatin. RITS requires siRNAs for its association with chromatin, and, based on its similarity to RISC, it seemed likely that RITS would also be targeted via base-pairing of siRNAs, either to DNA or RNA1. The dependence on transcription suggests that the target is RNA.
Schramke et al.2 show that the transcription of the gene targeted for silencing must be carried out specifically by RNApII. Although a polymerase from bacteriophage (a virus that infects only bacteria) can produce transcripts from a eukaryotic chromosome, silencing does not occur. Therefore, although RITS may be targeted via base-pairing to nascent RNA transcripts, an additional mechanism must exist for specifically coupling silencing to RNApII. The two groups2,3 find that very different mutations in RNApII disrupt the formation of heterochromatin: truncation of the RNApII largest-subunit carboxy-terminal domain (the CTD, normally required for coupling mRNA synthesis to mRNA processing4), or a specific point mutation in the RNApII Rpb2 subunit, both lead to loss of silencing, but with an interesting difference. Whereas siRNAs are made normally in the CTD truncation mutant, the Rpb2 mutant seems to be blocked in processing the siRNAs. Therefore, RNApII may have multiple roles in the siRNA pathway.
Perhaps the simplest model to explain the coupling of RNA-induced silencing with transcription (Fig. 1) is that RITS is tethered to some part of the RNApII elongation complex, which produces the target mRNA. Through this interaction, as well as recognition of specific transcript sequences, the histone methyltransferase would be localized to the appropriate target gene and so could modify the histones as the gene is being transcribed. Indeed, molecular-interaction experiments show that RITS is linked to nascent transcripts2,5 as well as to RNApII (ref. 2). Furthermore, RITS and its associated factors can be chemically crosslinked to genes undergoing silencing, and this requires siRNA and the nascent transcripts2,5. There is clear precedent for the coupling of transcription with chromatin modification: two other histone methyltransferases (Set1 and Set2) bind directly to elongating RNApII and modify transcribed regions of genes6.
However, there are several other models that could explain why only RNApII can mediate RITS-dependent silencing. It is possible that RITS interacts with the transcript not only through base-pairing, but also by recognizing an RNApII-specific modification of mRNA (such as the cap structure or poly(A) tail). In this regard, it is interesting that a screen for factors that promote RNAi in the nematode worm Caenorhabditis elegans identified several factors that are required for proper mRNA processing7. The mechanisms by which these factors affect RNAi are unknown.
In addition to mRNA-processing enzymes, the RNApII elongation complex carries several chromatin-modifying enzymes6. Although bacteriophage polymerases and eukaryotic RNA polymerase III can transcribe through chromatin without disrupting nucleosomes8, passage of the RNApII elongation complex leads to large changes in the chromatin9. Histone subunits may be exchanged as the transcription complex passes through a nucleosome. Furthermore, several transcription-dependent modifications of histones, including the methylations described above, have been identified. It may be that one or more of these alterations are prerequisites for the RITS-associated histone methyltransferase to modify its substrate.
Each of these explanations makes some testable predictions, so the process that links RITS-mediated repression specifically to RNApII may soon become clear. However, other questions remain. For example, is transcription required only to initiate silencing, with the repressive chromatin being subsequently propagated by epigenetic mechanisms such as histone methylation? Or is some low level of transcription paradoxically required to maintain repression? Any transcripts made from a ‘silenced’ gene would be subject to RISC-mediated degradation, so some leakiness of transcriptional silencing may easily be tolerated. An RNA polymerase IV that apparently specializes in synthesizing siRNA precursors has recently been discovered in plants10. This suggests another mechanism for simultaneous transcription and silencing: perhaps RNApIV can transcribe through chromatin structures that block RNApII. It will be interesting to see whether transcription by RNApIV is directly coupled to the silencing complex.
Another question is whether siRNAs are themselves generated during transcription, a possibility suggested by the findings of Kato et al.3. Once targeted by RISC/RITS, the transcript slated for destruction can be used to generate new siRNAs. This requires two enzymes — an RNA-dependent RNA polymerase that generates double-stranded RNA, and the Dicer enzyme that cuts the RNA into siRNA lengths. Both enzymes associate with RITS or RISC complexes and could therefore be present at the site of transcription.
One of the main functions of the RNAi system is probably to suppress expression of the repetitive elements that parasitize eukaryotic genomes. A two-pronged approach to silencing makes good sense. The RISC complex can target any transcripts that manage to reach the cytoplasm from the nucleus, preventing them from being translated. But if this were the only mechanism, considerable cellular energy might still be wasted in making transcripts from the repetitive elements. The nuclear RITS complex can repress expression of those RNAs before they are even made. Because target recognition uses complementary RNA sequences, once a particular element or gene is recognized by the RNAi system, all copies in the cell will be targets for inactivation. Nucleotide sequences provide a high level of specificity, but there must be an opportunity for the target sequences to be recognized. By coupling the RNAi machinery to ongoing transcription, siRNAs can identify target transcripts as they are synthesized, resulting in efficient and almost immediate repression.
Ekwall, K. Mol. Cell 13, 304–305 (2004).
Schramke, V. et al. Nature 435, 1275–1279 (2005).
Kato, H. et al. Science published online 9 June 2005 (doi:10.1126/science.1114955).
Maniatis, T. & Reed, R. Nature 416, 499–506 (2002).
Motamedi, M. R. et al. Cell 119, 789–802 (2004).
Hampsey, M. & Reinberg, D. Cell 113, 429–432 (2003).
Kim, J. K. et al. Science 308, 1164–1167 (2005).
Studitsky, V. M., Kassavetis, G. A., Geiduschek, E. P. & Felsenfeld, G. Science 278, 1960–1963 (1997).
Ahmad, K. & Henikoff, S. Cell 111, 281–284 (2002).
Vaughn, M. W. & Martienssen, R. A. Mol. Cell 17, 754–756 (2005).
About this article
Mass Spectrometry Reviews (2011)
Recognition of Polyadenosine RNA by the Zinc Finger Domain of Nuclear Poly(A) RNA-binding Protein 2 (Nab2) Is Required for Correct mRNA 3?-End Formation*
Journal of Biological Chemistry (2010)
Journal of Biological Chemistry (2010)
The Plant Journal (2008)
Evaluation of anti-Fas ligand-induced apoptosis and neural differentiation of PC12 cells treated with nerve growth factor using small interfering RNA method and sampling by microdialysis
Analytical Biochemistry (2007)