The discovery that small interfering RNA duplexes (siRNA) can silence gene expression in mammalian cells has revolutionized biomedical research. The most successful application of the discovery has been to study gene function in cultured human or mouse cells. However, the knockdown effect of siRNA is only transient. To achieve a more sustained gene-silencing effect, shRNA (small hairpin RNA) expressed from a vector is preferred. An additional benefit of shRNA is that RNA interference (RNAi) can now be applied in vivo through delivering shRNA-expressing vectors by transgenic technology. Transgenic RNAi not only allows the study of biological processes not present in cultured cells but also offers chronic therapeutic potentials. In this review, we will summarize the developments in the generation of transgenic RNAi mice.
A gene’s biological role in an organism can only be fully appreciated by observing the phenotypic consequences of altering its function. The approaches to alter a gene’s function in the model organism, mouse, are transgenesis and targeted gene disruption. In the transgenic approach, one expresses a gene (transgene) at a different time or place, or at a different (usually higher) level than the endogenous gene. Then one asks what effect the transgene has on the mouse, developmentally and/or pathophysiologically. This is a gain of function type of experiment, meaning the gene has gained some additional functions (changed time, place, and level of expression). The most revealing experiment, however, is to observe the effect of the loss of function of a gene. In mice, the standard approach to achieve loss of function is to disrupt the locus of interest through the embryonic stem (ES) cell technology.
The targeted gene disruption (or knockout) approach has been nearly the only method available to inactivate a gene in mammals, until the advent of RNA interference (RNAi). RNAi induced by short double-stranded RNA (dsRNA) has gained widespread application in biological research since its discovery (7, 14, 15). The discovery of RNAi was rooted in the desire to alter gene function by an antisense approach in C. elegans. It was observed that introducing dsRNA homologous to a specific gene resulted in the posttranscriptional silencing of that gene (9). This dsRNA-induced gene silencing was termed RNAi. It occurs through two main steps. The dsRNA is initially recognized by an enzyme of the RNase III family of nucleases, named Dicer, and processed into small double-stranded molecules, termed siRNA (small interfering RNA) (27). siRNA is then incorporated into a large multi-subunit nuclease complex RISC (RNA-induced silencing complex). RISC unwinds the siRNA in an ATP-dependent step and finds homologous target mRNAs using the siRNA sequence as a guide and cleaves these mRNAs (22, 41).
The siRNA-mediated gene silencing is likely a self-defense mechanism against viral invasions. Animal cells also express a class of short (~22 nt) RNAs termed micro-RNA (miRNA) (19–21). miRNAs are initially transcribed as long primary miRNA (pri-miRNA) precursors most likely by RNA polymerase II (Pol II) and are processed by Drosha and Dicer nucleases (1). They regulate gene expression through blocking the translation of the target mRNAs or inducing the degradation of the targets just as siRNA does (1).
The ability of dsRNA to alter gene expression is well known in mammals. As part of the antiviral defense, an interferon-inducible protein kinase named PKR is activated by dsRNA intermediates that resulted from the viral replication. PKR contains two dsRNA binding domains, and its activity depends on dsRNA (23). Among many cellular targets of PKR is the small subunit of the eukaryotic initiation factor 2 (eIF2α). Phosphorylation of eIF2α by PKR causes a generalized inhibition of translation. dsRNA also activates 2′,5′-oligoadenylate synthetase (2′,5′-AS) (24, 25). Activated 2′,5′-AS causes nonspecific mRNA degradation due to the activation of RNase L by oligoadenylate. The end result of PKR and 2′,5′-AS activation is global, non-specific gene silencing. Eventually, cells die via apoptosis (see Ref. 11 for a review). The key difference between dsRNA response in mammalian cells and RNAi is the specificity. The former is global and nonspecific, silencing all protein-encoding RNAs by inhibiting their translation and promoting their degradation, whereas the latter is specific to the target mRNA based on the homologous sequence in siRNA.
The global response to dsRNA in mammalian cells prohibited the use of dsRNA as a gene-silencing tool, until the discovery that synthetic short (21 nt) RNA duplexes could induce RNAi in cultured mammalian cells without eliciting the PKR reaction (7). This method was spread widely and quickly because it made possible, for the first time, that one could inactivate a gene with high efficiencies in cultured mammalian cells with different tissue origins. A substantial improvement of this methodology was achieved with the demonstration that the short RNA duplex could be produced in vivo via expressing vectors (viral and non-viral) to stably induce RNAi in mammalian cells (3, 22, 26, 29, 31, 44). The short RNA duplex is often expressed as small hairpin RNA (shRNA), which is believed to be processed to siRNA inside the cells (3). More recently, artificial miRNA-expressing vectors (shRNA-mir) have been developed in which shRNA sequences are engineered into the natural miR-30 RNA hairpin to achieve better knockdown efficiencies (37).
These technical advancements have made it possible to perform loss of function genetic studies in cultured cells and even genetic screenings using shRNA libraries (2, 30). However, there are limitations to the types of biological processes that can be studied in cultured cells. Many genes function in processes that cannot be recapitulated in cultured cells. Therefore, it is desirable to apply the RNAi technology in the whole mouse as an alternative to the targeted gene disruption approach in delineating gene function. FIGURE 1⇓ summarizes the RNAi process.
Transgenic mice have been used for the expression of “foreign” genes for decades and the technology to generate such mice has been well established. In theory, shRNA or shRNA-mir can be expressed in the same way as the other transgenes in mice to achieve gene-silencing effects. We therefore designed a transgenic shRNA vector (pTshRNA) (33) based on the original pSUPER (3), in which the shRNA expression is driven by H1 promoter (FIGURE 2⇓). In pTshRNA, the 3′ end (~1 kb) of the H1 gene was included, which may have helped the expression of shRNA. In addition, an EGFP cassette was added for easy genotyping of the transgenic animals. As a test, we generated two constructs, one against p57KIP2 and the other cyclin D1. The constructs were purified free of the plasmid backbone and injected into pronuclei to produce transgenic mice, which were analyzed before the birth (thus at the F0 generation) or used to establish transgenic lines. We demonstrated knockdown effects in a number of tissues and showed that the RNAi effect could be transmitted for many generations (33).
The shRNA expression construct could also be introduced into one-cell embryos through lentiviral infection or direct injection of the viruses (35). However, this method requires a high-titer viral stock, which is not easily obtained. Retroviral vectors will not work as transgenes since they will be silenced in mouse embryos.
Another way to generate transgenic RNAi mice is to introduce shRNA- or shRNA-mir-expressing constructs into mouse ES cells via electroporation or lentiviral infection (5) (FIGURE 3⇓). The knockdown effect can be first screened in the ES clones (if the target is not expressed in ES cells, one can introduce a expression construct to exogenously express the target). Then ES clones can be selected for desired knockdown efficiencies and used to generate mice just as any knockout ES clones. Kunath et al. (18) used this method to produce mice with Rasa1 (also known as RasGAP) knockdown. They generated a simple vector in which the shRNA sequence was placed downstream of H1 promoter. Twelve ES cell lines were established, and 11 of them had significant reduction in RasGAP expression. Embryos derived solely from the selected ES clones were produced through the tetraploid aggregation technique (8, 32). All of the Rasa1 knockdown embryos displayed phenotypes closely resembling those in Rasa1 knockout mice (18).
Conditional RNAi in vivo
To better analyze gene function, especially the essential ones, it is desirable to express RNAi effect in vivo in a controlled manner, both temporally and spatially. Various techniques in conditional knockout approaches can be modified for the use in conditional knockdowns. Cre-loxP system has been utilized to conditionally activate or inactivate RNAi in an irreversible manner (6, 16, 17, 42, 43). In the activation system, a loxP-flanked random stuffer sequence is inserted between the Pol III promoter and the shRNA-encoding sequence. On the action of Cre recombinase, the recombination of the two loxP sites excises the stuffer sequence, allowing shRNA expression (FIGURE 4A⇓). In the inactivation system, a Pol III-shRNA expression cassette is flanked by loxP sites and can be removed by Cre excision, thus leading to the termination of knockdown (FIGURE 4B⇓). One can take advantage of the availability of various transgenic mouse lines that express Cre conditionally to generate spatiotemporally controlled RNAi effect in vivo.
Furthermore, reversible gene knockdown can be accomplished through controlling the expression of RNAi triggers, either shRNA or shRNA-mir, with a drug-inducible promoter. One method that has been reported to be effective in vivo is the tet-KRAB-based lentiviral vector-derived system (40). This system utilizes the KRAB domain found in many zinc-finger proteins, which can silence both Pol II and Pol III promoters by triggering heterochromatin formation. When fused to the tetR DNA binding domain, the resulting chimeric protein can bind to the tetO site in the presence or absence of doxycycline, depending on the tetR version used. In this way, any promoter adjacent to the tetO site will be silenced, preventing the expression of RNAi trigger sequences.
Moreover, any characterized Pol II promoter can be used to express shRNA or shRNA-mir in a tissue-specific manner. Rao et al. (34) used the proximal promoter from the mouse Pem (Rhox5) gene to express shRNA-mir against WT1 (Wilms’ tumor 1) in Sertoli cells. The resulted transgenic mice show reduced expression of WT1 in Sertoli cells of the adult testes, and as a consequence of the reduced expression, spermatogonia display increased apoptotic death. Since the lack of WT1 causes embryonic lethality, the requirement of WT1 in Sertoli cells was unknown. Thus the transgenic RNAi approach nicely complements the knockout studies.
Genetic screening via transgenic RNAi
The beauty of transgenic RNAi is that, unlike gene knockouts, one does not have to breed the transgene into homozygosity to see the effect. In fact, one can even observe phenotypic consequences of the RNAi effect on trans-genic embryos or mice at the F0 generation. Thus shRNA transgenes behave like dominant-negative alleles of the genes of interest. This feature, combined with the relatively inexpensive pronuclear injection technique, makes it possible to perform genetic screenings in mice. For developmental phenotypes, the screening can be done with little mouse cage costs since there is no breeding needed when the injected one-cell embryos can be scored directly after they have developed to certain stages. Obviously, the screening cannot be as high throughput as in lower organisms such as C. elegans (10, 38), since the constructs have to be injected one by one. However, it is high throughput comparing with other mutagenesis strategies in mice. In a pilot experiment, we screened a group of 20 genes for their role in the development of the kidney. These genes were chosen based on their expression profile in the kidney. Two pTshRNA constructs were made for each gene and injected into one-cell mouse embryos. The injected embryos were harvested 18 days later, and the transgenic embryos were identified by their expression of GFP carried on pTshRNA. Even though the scale of this pilot screening was small, we did identify one gene, Id4, which may play a role in the maturation of the medulla of the kidney (33).
RNAi in adult mice
Both siRNAs (RNA duplexes) and shRNAs (expressing constructs in the form of plasmids) can be delivered directly into adult animals via various strategies. In these experiments, the RNAi effect is confined to part of the body, often transient, and will not transmit to the next generation. More importantly, due to the variations in the knockdown effect among individuals, a large number of animals are required to obtain statistically significant results. The most commonly employed delivery method is the hydrodynamic intravenous tail injection (12). In this method, the rapid injection of large volumes of nucleic acids increases venous pressure, allowing siRNA duplexes or shRNA-expressing vector plasmids to enter the organ. This method is most effective for the liver. Other delivery options include intraperitoneal injection (39) and topical administration (4). There are also reports on the successful delivery of siRNA using ultrasound, electropora-tion, and particle based “gene-gun” (for a review on delivery strategies, see Ref. 13). Viral shRNA constructs (most often adenoviruses) can also be delivered locally through injection into specific tissues (5a).
The methods mentioned above offer quick turnarounds in experiments, but the knockdown efficiencies vary greatly, and almost certainly not every cell in a target tissue or organ will be “transfected.” Thus transgenic RNAi through the germline is more preferable, despite that more time is needed. However, the hematopoietic system is an exception because of the well established methodologies to isolate hematopoietic stem cells and to reconstitute all blood cell lineages with the stem cells in irradiated recipient mice. One could establish RNAi in the stem cells first and use the treated stem cells to do the reconstitution, therefore being able to study a gene’s function in hematopoeisis (28).
A summary of various methods to generate RNAi effects in mice is presented in Table 1⇓. Although the use of transgenic RNAi mice is on the rise, it is unlikely that transgenic RNAi will replace conventional targeted gene disruption approach, because there are a number of limitations. First, RNAi-mediated knockdown of expression is never complete. Some residual expression of an RNAi-targeted gene is expected. Second, except modulating the level of expression, no other manipulations on a gene (for example, point mutations) can be achieved with RNAi. Third, it remains to be determined whether RNAi is also effective against miRNA, whereas the targeted gene disruption approach can remove or modify almost any genetic elements in the genome. Fourth, not all cell types in the body are competent to carry out RNAi, and there are huge differences in the knockdown efficiencies among those cell types that can perform RNAi (33). Finally, whether the transgenic RNAi effect is permanent is still an open question, although we have seen the effect up to six generations. The shRNA transgenes could be shut down developmentally or turned off through epigenetic modifications. However, transgenic RNAi does offer a unique feature not easily achievable through conventional gene knockout, that is, to reduce the expression of a gene to various degrees with different trans-genic lines so that one can study a gene’s function through a spectrum of expression levels. Furthermore, some residual expression of a gene after the knockdown may prove to be advantageous, for example, overcoming the lethality associated with a null mutation. Thus transgenic RNAi will not replace gene knockouts but can be a very effective complementary approach.
This work is supported by National Institutes of Health Grants EY-014745 and DK-063964 to P. Zhang.
- © 2007 Int. Union Physiol. Sci./Am. Physiol. Soc.