Short Technical Reports Wendel. Genes duplicated by polyploidy show unequal contributions to the transcrip- RNA Interference by tome and silencing that are reciprocal and or- Production of Short Hair- gan-specific. Proc. Natl. Acad. Sci. USA (In Press.) pin dsRNA in ES Cells, Their Differentiated The authors contributed equally to the application of this method and manuscript Derivatives, and in preparation. We thank Ines Alvarez, Bao Somatic Cell Lines Liu, Ryan Percifield, and David Senchina for assistance in evaluating the loci select- BioTechniques 34:734-744 (April 2003) ed, and especially Jonathan Wendel for pro- viding a stimulating laboratory environ- ment to develop this technique. This work was supported by a U.S. Department of ABSTRACT Agriculture (USDA) National Research Ini- tiative Grant and the Iowa State University dsRNA of several hundred nucleotides in Plant Sciences Institute (to J. Wendel) and a length is effective at interfering with gene ex- grant from the Jane Coffin Childs Memorial pression in mouse oocytes, pre-implantation Fund for Medical Research (to K.A.). Trade embryos, and embryonic stem (ES) cells but names and commercial products are men- is not as efficient in differentiated cell lines. tioned solely for information and no en- Here we describe a method to achieve RNA dorsement by the USDA is implied. Address interference in totipotent and differentiated correspondence to Dr. Richard Cronn, Pa- ES cells together with a wide range of other cific Northwest Research Station, 3200 SW mammalian cell types that is both simple and Jefferson Way, Corvallis, OR, 97331, USA. efficient. It utilizes a linearized plasmid that e-mail: [email protected]. directs the expression of a hairpin RNA with a 22-nucleotide-paired region. This molecule Received 31 October 2002; accepted has a 13-nucleotide 5′ overhang that would 15 January 2003. be subject to capping on its 5′ phosphoryl group and thus differs from the ideal struc- Richard C. Cronn and Keith ture suggested for effective small interfering L. Adams1 RNAs. Thus, it appears either that the struc- ture of small inhibitory RNA molecules may Pacific Northwest Research not need to be as precise as previously Station thought or that such a transcript is efficiently U.S. Forest Service processed to a form that is effective in inter- Corvallis, OR fering with gene expression. 1Iowa State University Ames, IA, USA INTRODUCTION The specific blocking gene expres- sion by double-stranded RNA (dsRNA) of up to several hundred nucleotides in length has been documented in a vari- ety of organisms including Caenorhab- ditis elegans (1), Drosophila melano- gaster (2), trypanosomes (3), planaria (4), paramecium (5), hydra (6), ze- brafish (7,8), and mouse (9,10). The use of dsRNA of such molecular weight for silencing gene expression has not been as efficient in cultured mammalian somatic cell lines where it appears to also inhibit gene expression nonspecifically (11). This reflects the ability of dsRNA to trigger the interfer- on response in mammalian cells, lead- Vol. 34, No. 4 (2003) Short Technical Reports ing to the activation of the RNA-depen- Moreover, when transfected into cul- and efficient method for the inhibition dent protein kinase PKR and 2′,5′- tured mammalian cell lines, small inter- of gene expression in mouse ES cells, oligoadenylate synthetase (2′,5′-AS). fering RNAs (siRNAs) of 23 nu- their differentiated derivatives, and in Activated PKR stalls translation by cleotides appear to evade the interferon somatic cell lines, by producing phosphorylation of the initiation factor response to elicit specific gene silenc- siRNAs in vivo as short hairpin struc- eIF2, and activated 2′,5′-AS causes ing (24,25). Recently, plasmids that di- tures transcribed from an RNA poly- mRNA degradation by 2′,5′-oligoad- rect the synthesis of a small hairpin merase II promoter. These RNAs are enylate-activated ribonuclease L (12). RNA from an RNA polymerase III pro- most likely not polyadenylated, so they These responses to introduction of any moter have been described (26–28). may be an effective trigger for the dsRNA lead to an overall block of The products are analogous to synthetic RNAi response. The effectiveness of translation and cell death. However, siRNAs because they have 3′ over- these transcripts suggests that either the three reports have now suggested such hanging T or U nucleotides but are precise structure of siRNA molecules longer dsRNAs can mediate RNA in- folded back on themselves to give a 19- may not be as critical as previously terference (RNAi) without accompany- to 20-nucleotide base paired region and thought (25) or that the structure of the ing interferon responses (13–15). a 4- to 9-nucleotide loop. transcript per se facilitates its conver- Longer dsRNA appears to mediate In this study, we describe a simple sion into effective siRNA. specific RNAi efficiently in mouse oocytes and embryos (9,10,16). Simi- larly, embryonic stem (ES) cells, which are derived from the mouse embryo, seem to show a specific response to RNAi with dsRNA of this size (17,18). These findings suggest that either em- bryonic cells have a more effective RNAi response or that the interferon response to dsRNA is absent or not yet fully developed. dsRNA with an average length of 500 bp has been typically delivered by microinjection into embryos or syncy- tial tissues or by transfecting cultured cells with dsRNA. The efficiency of this approach has recently been im- proved by an electroporation approach that can introduce dsRNA into multiple mouse oocytes or embryos (16). In C. elegans, RNAi is sufficiently robust that the effects can be achieved by feeding the worms an E. coli strain that is expressing the specific dsRNA (19). Interference with specific gene expres- sion can also be achieved in stable transgenic lines of D. melanogaster and C. elegans by the expression of a con- struct directing the production of a hair- pin loop of RNA (20,21). Such con- structs can be placed under the control of regulatable promoters to interfere with specific gene expression in a spa- tiotemporal manner (21,22). An important clue to understanding the mechanism of RNAi came from the discovery that 21–23 nucleotide pieces Figure 1. Schematic view of the construct encoding for short hairpin GFP RNA. (A) Linearized pds- of dsRNA are intermediates that are GFP-N1 plasmid. Double-stranded synthetic DNA fragment was inserted between NheI and XbaI sites of generated from the cleavage of longer the pGFPesm-N1 plasmid, downstream of the CMV promoter. (B) The inserted synthetic DNA fragment dsRNA (23). Such small dsRNAs consisted of indirect repeats of a 22-nucleotide-long GFP sequence that were separated by an 18-nu- cleotide-long spacer of three restriction sites. Flanking restriction sites were added to the insert to pro- induce efficient sequence-specific duce a NheI site at its 5′ end and a XbaI site at its 3′ end. (C) Predicted structure of the short hairpin RNA: mRNA degradation in lysates prepared a 7-MeG cap is likely to be added to the 5′ ends of a large proportion of the transcripts. The 5′-transcribed from Drosophila embryos (23,24). sequence preceding the NheI restriction site is marked in red and the target GFP sequence in green. 736 BioTechniques Vol. 34, No. 4 (2003) MATERIALS AND METHODS nealed DNA was cloned between the nonessential amino acids (Invitrogen). NheI and XbaI sites of the plasmid ES cells used for experiments were kept Plasmid for dsRNA Production pGFPemdN1 such that the inverted re- in culture between 15 and 25 passages peat was located downstream from the from the isolation of ES cells from the Two oligonucleotides were synthe- cytomegalovirus (CMV) promoter and inner cell mass of the blastocyst. To in- sized corresponding to an inverted se- upstream of the XbaI restriction site in duce differentiation, LIF was removed quence repeat of a 22-nucleotide frag- this construct (Figure 1). from the medium. The cells were trans- ment of GFP carried in the pQBIgfp fected 10 days after the induction of dif- vector (Qantum/Qbiogene, Montreal, Cell Culture ferentiation. QC, Canada), 5′-GTGGAGAGGGTG- STO cells were grown in modified AAGGTGATGC-3′, which extended Wild-type (CBB) ES cells (kindly DMEM (Invitrogen) supplemented from nucleotides 92 to 113, relative to provided by Azim Surani, University of with 10% FBS (Sigma, St. Louis, MO, the first nucleotide of the start codon. Cambridge) and transgenic (β-actin) ES USA). NIH-3T3 cells were grown in The inverted repeats were separated by cells expressing Mus musculis GFP DMEM supplemented with 10% FBS three unique restriction sites (SpeI- (MmGFP) under the β-actin promoter (both from Sigma). NotI-SalI), and additional unique re- (Zernicka-Goetz, unpublished data) striction sites were added to both ends. were grown on gelatinized plates in Transfection The two complementary oligonu- DMEM/nut mixture F12 (Invitrogen, cleotides were incubated at 80°C and Carlsbad, CA, USA) supplemented with Prior to transfection, the pdsGFP-N1 left until the temperature decreased to 12% FBS (Invitrogen), 100 IU/mL plasmid was linearized with XbaI. The 30°C. After annealing, the resulting ds- leukemia inhibitory factor (LIF) pCMV-nls LacZ plasmid was kindly DNA fragment possessed an NheI re- (CHEMICON, Temecula, CA, USA), provided by Dr. Aaron Zorn (Universi- striction site at the 5′ end and a XbaI L-glutamine, sodium bicarbonate, β- ty of Cambridge). Synthetic 21-bp restriction site at the 3′ end. The an- mercaptoethanol, sodium pyruvate, and siRNAs [kindly provided by J. Pearson Short Technical Reports (University of Cambridge)] correspond- amount of protein was used for the lumi- wild-type ES cells that expressed GFP ing to nucleotides 469–489 of QBIgfp nescent measurements. β-galactosidase transiently from an exogenous plasmid, and to nucleotides 489–509 of MmGFP expression was monitored with the Lu- pQBIgfp.