Post-transcriptional gene silencing (PTGS), which was initially considered a bizarre phenomenon limited to petunias and a few other plant species, is now one of the hottest topics in molecular biology (1). In the last few years, it has become clear that PTGS occurs in both plants and animals and has roles in viral defense and transposon silencing mechanisms. Perhaps most exciting, however, is the emerging use of PTGS and, in particular, RNA interference (RNAi) PTGS initiated by the introduction of double-stranded RNA (dsRNA) as a tool to knock out expression of specific genes in a variety of organisms (reviewed in 1-3).
How was RNAi discovered? How does it work? Perhaps more importantly, how can it be harnessed for functional genomics experiments? This article will briefly answer these questions and provide you with resources to find in depth information on PTGS and RNAi research.
A Bizarre Phenomenon is Discovered: Cosuppression and PTGS in Plants
More than a decade ago, a surprising observation was made in petunias. While trying to deepen the purple color of these flowers, Rich Jorgensen and colleagues introduced a pigment-producing gene under the control of a powerful promoter. Instead of the expected deep purple color, many of the flowers appeared variegated or even white. Jorgensen named the observed phenomenon "cosuppression", since the expression of both the introduced gene and the homologous endogenous gene was suppressed (1-5).
First thought to be a quirk of petunias, cosuppression has since been found to occur in many species of plants. It has also been observed in fungi, and has been particularly well characterized in Neurospora crassa, where it is known as "quelling" (1-3).
But what causes this gene silencing effect? Although transgene-induced silencing in some plants appears to involve gene-specific methylation (transcriptional gene silencing, or TGS), in others silencing occurs at the post-transcriptional level (post-transcriptional gene silencing, or PTGS). Nuclear run-on experiments in the latter case show that the homologous transcript is made, but that it is rapidly degraded in the cytoplasm and does not accumulate (1, 3, 6).
Introduction of transgenes can trigger PTGS, however silencing can also be induced by the introduction of certain viruses (2, 3). Once triggered, PTGS is mediated by a diffusible, trans-acting molecule. This was first demonstrated in Neurospora, when Cogoni and colleagues showed that gene silencing could be transferred between nuclei in heterokaryotic strains (1, 7). It was later confirmed in plants when Palauqui and colleagues induced PTGS in a host plant by grafting a silenced, transgene-containing source plant to an unsilenced host (8). From work done in nematodes and flies, we now know that the trans-acting factor responsible for PTGS in plants is dsRNA (1-3).
Gene Silencing by dsRNA: RNA Interference
RNAi Is Discovered in Nematodes
The first evidence that dsRNA could lead to gene silencing came from work in the nematode Caenorhabditis elegans. Seven years ago, researchers Guo and Kemphues were attempting to use antisense RNA to shut down expression of the par-1 gene in order to assess its function. As expected, injection of the antisense RNA disrupted expression of par-1, but quizzically, injection of the sense-strand control did too (9).
This result was a puzzle until three years later. It was then that Fire and Mello first injected dsRNA a mixture of both sense and antisense strands into C. elegans (10). This injection resulted in much more efficient silencing than injection of either the sense or the antisense strands alone. Indeed, injection of just a few molecules of dsRNA per cell was sufficient to completely silence the homologous gene's expression. Furthermore, injection of dsRNA into the gut of the worm caused gene silencing not only throughout the worm, but also in its first generation offspring (10).
The potency of RNAi inspired Fire and Timmons to try feeding nematodes bacteria that had been engineered to express dsRNA homologous to the C. elegans unc-22 gene. Surprisingly, these worms developed an unc-22 null-like phenotype (11-13). Further work showed that soaking worms in dsRNA was also able to induce silencing (14). These strategies, whereby large numbers of nematodes are exposed to dsRNA, have enabled large-scale screens to select for RNAi-defective C. elegans mutants and have led to large numbers of gene knockout studies within this organism (15-18).
RNAi in Drosophila
RNAi has also been observed in Drosophila. Although a strategy in which yeast were engineered to produce dsRNA and then fed to fruit flies failed to work, microinjecting Drosophila embryos with dsRNA does effect silencing (2). Silencing can also be induced by "shooting" dsRNA into Drosophila embryos with a "gene gun" or by engineering flies to carry DNA containing an inverted repeat of the gene to be silenced. Over the last few years, these RNAi strategies have been used as reverse genetics tools in Drosophila organisms, embryo lysates, and cells to characterize various loss-of-function phenotypes (2, 19-23).
GLOSSARY OF TERMS
Cosuppression - Silencing of an endogenous gene caused by the introduction of a transgene or infection by a virus. This term, which can refer to silencing at the post-transcriptional (PTGS) or transcriptional (TGS) level, has been primarily adopted by researchers working with plants.
Post-transcriptional Gene Silencing (PTGS) - Silencing of an endogenous gene caused by the introduction of a homologous dsRNA, transgene or virus. In PTGS, the transcript of the silenced gene is synthesized but does not accumulate because it is rapidly degraded. This is a more general term than RNAi, since it can be triggered by several different means.
Quelling - PTGS in Neurospora crassa induced by the introduction of a transgene.
RISC - RNA-induced silencing complex. A nuclease complex, composed of proteins and siRNA (see below), that targets and destroys endogenous mRNAs complementary to the siRNA within the complex.
RNA interference (RNAi) - Post-transcriptional gene silencing (PTGS) induced by the direct introduction of dsRNA. The term "RNA interference" was first used by researchers studying C. elegans.
siRNAs - Small interfering RNAs. Current models of PTGS indicate that these 21-23 nucleotide dsRNAs mediate PTGS. Introduction of siRNAs can induce PTGS in mammalian cells. siRNAs are apparently produced in vivo by cleavage of dsRNA introduced directly or via a transgene or virus. Amplification by an RNA-dependent RNA polymerase (RdRP) may occur in some organisms. siRNAs are incorporated into the RNA-induced silencing complex (RISC), guiding the complex to the homologous endogenous mRNA where the complex cleaves the transcript.
The Biochemical Mechanism of RNAi
So how does injection of dsRNA lead to gene silencing? Many research groups have diligently worked over the last few years to answer this important question. A key finding by Baulcombe and Hamilton provided the first clue. They identified RNAs of ~25 nucleotides in plants undergoing cosuppression that were absent in non-silenced plants. These RNAs were complementary to both the sense and antisense strands of the gene being silenced (24).
Further work in Drosophila using embryo lysates and an in vitro system derived from S2 cells shed more light on the subject (3, 25, 26). In one notable series of experiments, Zamore and colleagues found that dsRNA added to Drosophila embryo lysates was processed to 21-23 nucleotide species. They also found that the homologous endogenous mRNA was cleaved only in the region corresponding to the introduced dsRNA and that cleavage occurred at 21-23 nucleotide intervals (26). Rapidly, the mechanism of RNAi was becoming clear.
Current Models of the RNAi
Both biochemical and genetic approaches (see "The Genes and Enzymes Involved in PTGS and RNAi" below for a discussion of genetic approaches used to undersand RNAi) have led to the current models of the RNAi mechanism. In these models, RNAi includes both initiation and effector steps (27, see also a Flash animation of "How Does RNAi Work?", from reference 3).
In the initiation step, input dsRNA is digested into 21-23 nucleotide small interfering RNAs (siRNAs), which have also been called "guide RNAs" (reviewed in 3, 18, 27). Evidence indicates that siRNAs are produced when the enzyme Dicer, a member of the RNase III family of dsRNA-specific ribonucleases, processively cleaves dsRNA (introduced directly or via a transgene or virus) in an ATP-dependent, processive manner. Successive cleavage events degrade the RNA to 19-21 bp duplexes (siRNAs), each with 2-nucleotide 3' overhangs (27, 28).
In the effector step, the siRNA duplexes bind to a nuclease complex to form what is known as the RNA-induced silencing complex, or RISC. An ATP-depending unwinding of the siRNA duplex is required for activation of the RISC. The active RISC then targets the homologous transcript by base pairing interactions and cleaves the mRNA ~12 nucleotides from the 3' terminus of the siRNA (3, 18, 27, 29). Although the mechanism of cleavage is at this date unclear, research indicates that each RISC contains a single siRNA and an RNase that appears to be distinct from Dicer (27).
Because of the remarkable potency of RNAi in some organisms, an amplification step within the RNAi pathway has also been proposed. Amplification could occur by copying of the input dsRNAs, which would generate more siRNAs, or by replication of the siRNAs themselves (see "Possible Role for RNA-dependent RNA Polymerase" below). Alternatively or in addition, amplification could be effected by multiple turnover events of the RISC (3, 18, 27).
The Genes and Enzymes Involved in PTGS and RNAi
Possible Role for RNA-dependent RNA
Genetic screens in Neurospora, C. elegans, and Arabidopsis have identified several genes that appear to be crucial for PTGS and RNAi. Several of these, including Neurospora qde-1, Arabidopsis SDE-1/SGS-2 and C. elegans ego-1, appear to encode RNA-dependent RNA polymerases (RdRPs). At first glance, it might be assumed that this is proof that an RdRP activity is required for RNAi. Certainly the existence of an RdRP might explain the remarkable efficiency of dsRNA-induced silencing if it amplifed either the dsRNA prior to cleavage or the siRNAs directly. But mutants of these genes have varying phenotypes, which makes the role of RdRP in RNAi difficult to discern (1, 3, 17, 18).
In C. elegans ego-1 mutants ("ego" stands for "enhancer of glp-1"), RNAi functions normally in somatic cells, but is defective in germline cells where ego-1 is primarily expressed. In Arabidopsis SDE-1/SGS-2 mutants ("SGS" stands for suppressor of gene silencing), siRNAs are produced when dsRNA is introduced via an endogenously replicating RNA virus, but not when introduced by a transgene. It has been proposed that perhaps the viral RdRP is substituting for the Arabidopsis enzyme in these mutants (1, 3, 17, 18). Although no homolog of an RdRP has been found in flies or humans, an RdRP activity has recently been reported in Drosophila embryo lysates (30). One model of amplification, termed the "random degradative PCR" model, suggests that an RdRP uses the guide strand of an siRNA as a primer for the target mRNA, generating a dsRNA substrate for Dicer and thus more siRNAs (27, 30). Evidence supporting this model has been found in worms, whereas experimental results refuting the model have been obtained from Drosophila embryo lysates (26, 27).
Two C. elegans genes, rde-1 and rde-4 ("rde" stands for "RNAi deficient"), are believed to be involved in the initiation step of RNAi. Mutants of these genes produce animals that are resistant to silencing by injection of dsRNA, but silencing can be effected in these animals by the transmission of siRNA from heterozygous parents that are not silencing deficient. The C. elegans rde-1 gene is a member of a large family of genes and is homologous to the Neurospora qde-2 ("qde" stands for "quelling deficient") and the Arabidopsis AGO1 genes ("AGO" stands for "argonaute"; AGO1 was previously identified to be involved in Arabidopsis development). Although the function of these genes in PTGS is unclear, a mammalian member of the RDE-1 family has been identified as a translation initiation factor. Interestingly, Arabidopsis mutants of AGO1, which are defective for cosuppression, also exhibit defects in leaf development. Thus some processes or enzymes involved in PTGS may also be involved in development (1, 3, 17, 18).
Important genes for the effector step of PTGS include the C. elegans rde-2 and mut-7 genes. These genes were initially identified from heterozygous mutant worms that were unable to transmit RNAi to their homozygous offspring (16). Worms with mutated rde-2 or mut-7 genes exhibit defective RNAi, but interestingly, they also demonstrate increased levels of transposon activity. Thus, silencing of transposons appears to occur by a mechanism related to RNAi and PTGS. Although the rde-2 gene product has not yet been identified, the mut-7 gene encodes a protein with homology to the nuclease domains of RNase D and a protein implicated in Werner syndrome (a rapid aging disease) in humans (1, 3, 17, 18, 31). Perhaps this protein is a candidate for the nuclease activity required for target RNA degradation.
PTGS Has Ancient Roots
Discoveries from both genetic and biochemical approaches point to the fact that PTGS has deep evolutionary roots. Proposals have been put forth that PTGS evolved as a defense mechanism against transposons or RNA viruses, perhaps before plants and animals diverged (1, 3, 17, 18).
Interestingly, it was noted by many researchers that disruption of genes required for RNAi often causes severe developmental defects. This observation suggested a link between RNAi and at least one developmental pathway.
A group of small RNA molecules, known as small temporal RNAs (stRNAs), regulates C. elegans developmental timing through translational repression of target transcripts. Research indicates that the C. elegans lin-4 and let-7 stRNAs are generated from 70-nt transcripts following the folding of these longer transcripts into a stem-loop structure. The folded RNA molecules are cleaved to produce 22-nt stRNAs by the enzyme Dicer (called DCR-1 in C. elegans). Thus Dicer generates both siRNAs and stRNAs, and represents an intersection point for the RNAi and stRNA pathways (32-34).
Recently, nearly 100 additional ~22 nt RNA molecules, termed microRNAs (miRNAs), were identified in Drosophila, C. elegans, and HeLa cells (35-38). Much like lin-4 and let-7, these miRNAs are formed from precursor RNA molecules that fold into a stem-loop secondary structure. The newly discovered ~22 nt miRNAs are believed to play a role in regulation of gene expression, and at least two of them are known to require Dicer for their production (37). It appears that the use of small RNAs for both gene regulation and RNAi is a common theme throughout evolution.
Inducing RNAi in Mammalian Cells From Mechanism to Application
Non-specific Gene Silencing by Long
While the natural presence of RNAi had been observed in a variety of organisms (plants, protozoa, insects, and nematodes), evidence for the existence of RNAi in mammalian cells took longer to establish. Transfection of long dsRNA molecules (>30 nt) into most mammalian cells causes nonspecific suppression of gene expression, as opposed to the gene-specific suppression seen in other organisms. This suppression has been attributed to an antiviral response, which takes place through one of two pathways.
In one pathway, long dsRNAs activate a protein kinase, PKR. Activated PKR, in turn phoshorylates and inactivates the translation initiation factor, eIF2a, leading to repression of translation. (39) In the other pathway, long dsRNAs activate RNase L, which leads to nonspecific RNA degradation (40).
A number of groups have shown that the dsRNA-induced antiviral response is absent from mouse embryonic stem (ES) cells and at least one cell line of embryonic origin. (41, 42) It is therefore possible to use long dsRNAs to silence specific genes in these specific mammalian cells. However, the antiviral response precludes the use of long dsRNAs to induce RNAi in most other mammalian cell types.
siRNAs Bypass the Antiviral
Interestingly, dsRNAs less than 30 nt in length do not activate the PKR kinase pathway. This observation, as well as knowledge that long dsRNAs are cleaved to form siRNAs in worms and flies and that siRNAs can induce RNAi in Drosophila embryo lysates, prompted researchers to test whether introduction of siRNAs could induce gene-specific silencing in mammalian cells (43). Indeed, siRNAs introduced by transient transfection were found to effectively induce RNAi in mammalian cultured cells in a sequence-specific manner. The effectiveness of siRNAs varies the most potent siRNAs result in >90% reduction in target RNA and protein levels (44-46). The most effective siRNAs turn out to be 21 nt dsRNAs with 2 nt 3' overhangs. Sequence specificity of siRNA is very stringent, as single base pair mismatches between the siRNA and its target mRNA dramatically reduce silencing (44, 47). Unfortunately, not all siRNAs with these characteristics are effective. The reasons for this are unclear but may be a result of positional effects (46, 48, 49). For current recommendations on designing siRNAs, see "siRNA Design".
RNAi as a Tool for Functional Genomics
Although the history and mechanism of RNAi and PTGS are fascinating, many researchers are most excited about RNAi's potential use as a functional genomics tool. Already RNAi has been used to ascertain the function of many genes in Drosophila, C. elegans, and several species of plants. With the knowledge that RNAi can be induced in mammalian cells by the transfection of siRNAs, many more researchers are beginning to use RNAi as a tool in human, mouse and other mammalian cell culture systems.
In early experiments with mammalian cells, the siRNAs were synthesized chemically (Ambion is one of several companies that offer custom siRNA synthesis). Recently, Ambion introduced a kit (the Silencer siRNA Construction Kit) to produce siRNAs by in vitro transcription, which is a less expensive alternative to chemical synthesis, particularly when multiple different siRNAs need to be synthesized. Once made, the siRNAs are introduced into cells via transient transfection. Due to differences in efficacy, most researchers will synthesize 34 siRNAs to a target gene and perform pilot experiments to determine the most effective one. Transient silencing of more than 90% has been observed with this type of approach (44-46, 48, 49).
So far, injection and transfection of dsRNA into cells and organisms have been the main method of delivery of siRNA. And while the silencing effect lasts for several days and does appear to be transferred to daughter cells, it does eventually diminish. Recently, however, a number of groups have developed expression vectors to continually express siRNAs in transiently and stably transfected mammalian cells (50-56). Some of these vectors have been engineered to express small hairpin RNAs (shRNAs), which get processed in vivo into siRNAs-like molecules capable of carrying out gene-specific silencing (50, 53, 54, 56). The vectors contain the shRNA sequence between a polymerase III (pol III) promoter and a 4-5 thymidine transcription termination site. The transcript is terminated at position 2 of the termination site (pol III transcripts naturally lack poly(A) tails) and then folds into a stem-loop structure with 3' UU-overhangs. The ends of the shRNAs are processed in vivo, converting the shRNAs into ~21 nt siRNA-like molecules, which in turn initiate RNAi (50). This latter finding correlates with recent experiments in C. elegans, Drosophila, plants and Trypanosomes, where RNAi has been induced by an RNA molecule that folds into a stem-loop structure (reviewed in 3).
Another siRNA expression vector developed by a different research group encodes the sense and antisense siRNA strands under control of separate pol III promoters (52). The siRNA strands from this vector, like the shRNAs of the other vectors, have 5 thymidine termination signals. Silencing efficacy by both types of expression vectors was comparable to that induced by transiently transfecting siRNA.
The recent studies on RNAi have taken the research world by storm. The ability to quickly and easily create loss-of-function phenotypes has researchers rushing to learn as much as they can about RNAi and the characteristics of effective siRNAs. In the future, RNAi may even hold promise for development of gene-specific therapeutics. Much has been learned about this powerful technique, but additional information becomes available on an almost daily basis (see The RNA Interference Resource to learn about the very latest RNAi research and tools). It is not an understatement to say that the field of functional genomics is being revolutionized by RNAi.
Flash Animation: How Does RNAi Work?
Hammond, S.M., Caudy, A.A., Hannon, G.J. (2001) Post-transcriptional Gene Silencing by Double-stranded RNA. Nature Rev Gen 2: 110-119.