An approach to genomewide screens of expressed small interfering RNAs in mammalian cells

Abstract

To facilitate the construction of large genomewide libraries of small interfering RNAs (siRNAs), we have developed a dual promoter system (pDual) in which a synthetic DNA encoding a gene-specific siRNA sequence is inserted between two different opposing polymerase III promoters, the mouse U6 and human H1 promoters. Upon transfection into mammalian cells, the sense and antisense strands of the duplex are transcribed by these two opposing promoters from the same template, resulting in a siRNA duplex with a uridine overhang on each 3′ terminus. A single-step PCR protocol has been developed by using this dual promoter system that allows the production of siRNA expression cassettes in a high-throughput manner. We have shown that siRNAs transcribed by either the dual promoter vector or siRNA expression cassettes can induce strong and gene-specific suppression of both endogenous genes and ectopically expressed genes in mammalian cells. Furthermore, we have constructed an arrayed siRNA expression cassette library that targets >8,000 genes with two siRNA sequences per gene. A high-throughput screen of this library has revealed both known and unique genes involved in the NF-κB signaling pathway.

RNA interference (RNAi) is an evolutionarily conserved phenomenon in which gene expression is suppressed by the introduction of homologous double-stranded RNAs (dsRNAs). After dsRNA molecules are delivered to the cytoplasm of a cell, they are cleaved by the RNase III-like enzyme, Dicer, to 21- to 23-nt small interfering RNAs (siRNAs) (1). These siRNAs are then incorporated into a protein complex, the RNA-induced silence complex (RISC). The antisense strand of the duplex siRNA guides the RISC to the homologous mRNA, where the RISC-associated endoribonuclease cleaves the target mRNA, resulting in silencing of the target gene (2). RNAi has been successfully used to suppress gene expression in a variety of organisms including zebrafish, Caenorhabditis elegans, Drosophila, planaria, mice, and mammalian cells (3, 4). In C. elegans and Drosophila, RNAi is typically induced by the introduction of a long dsRNA (up to 1–2 kb) produced by in vitro transcription. This simple approach cannot be used in mammalian cells, where introduction of long dsRNA elicits a strong antiviral response that obscures any gene-specific silencing effect. Much of this response is caused by activation of the dsRNA-dependent protein kinase PKR, which phosphorylates and inactivates the translation initiation factor eIF2a (5, 6). However, introduction of 21-nt siRNAs into mammalian cells does not stimulate the antiviral response in mammalian cells and can effectively target specific mRNAs, resulting in gene silencing (7).

Short siRNA molecules can be prepared by chemical synthesis or in vitro transcription. Alternatively, they can be transcribed in vivo by using siRNA expression vectors with a RNA polymerase (pol) III promoter (including U6, human H1, and tRNA promoters) (8, 9), or a pol II promoter with a minimal poly(A) signal sequence (10). Tissue-specific pol II promoters can be used to carry out tissue-specific gene suppression. Typically a single promoter is used to express a short hairpin sequence, although two tandem pol III promoters have also been used to transcribe the sense and antisense siRNA sequences. In addition to plasmid-based systems, PCR-derived siRNA expression cassettes based on the single-promoter system have been shown to efficiently suppress transfected gene activity (11).

Genomewide RNAi screens in C. elegans using libraries of in vitro-transcribed long dsRNAs have proven extremely useful in gene discovery and functional annotation in various processes, including early embryonic development, lethality, sterility, genome instability, and longevity (12–15). Similarly, such approaches have also been carried out in cultured Drosophila cells for identification of Hedgehog pathway components (16). To date, the use of RNAi libraries in mammalian systems has been largely limited to specific protein families because of technical and practical issues associated with generating large synthetic siRNA and/or vector-based short hairpin RNA (shRNA) expression libraries. For example, using a small shRNA expression library focused on the family of deubiquitinating enzymes, Trompouki et al. (17) recently showed that the tumor suppressor, CYLD, negatively regulates NF-κB signaling by deubiqutination. In a similar fashion, a siRNA library targeted against kinases was used to identify modulators of Trail-induced apoptosis (18).

Herein, we report the development of a dual promoter siRNA expression system that allows the facile construction of siRNA expression libraries for genomewide screens. In this system, a gene-specific siRNA sequence is inserted between two different opposing pol III promoters, the mouse U6 and human H1 promoters. Upon transfection into mammalian cells, the sense and antisense strands of the siRNA duplex are transcribed by these two opposing promoters from the same template, resulting in a siRNA duplex with a uridine overhang on each 3′ terminus, similar to the siRNA generated by Dicer. These siRNAs can be incorporated into the RNA-induced silence complex without any further modification. Furthermore, a single-step PCR protocol has been developed that allows the production of siRNA expression cassettes in a high-throughput manner. We have used this methodology to construct a siRNA expression cassette library that targets >8,000 human genes with two designed sequences per gene. To demonstrate the utility of this library, it was screened in a pNF-κB luciferase (Luc) reporter assay, and both known and unique regulators of NF-κB signaling were identified.

Materials and Methods

Construction of Plasmids for siRNA Synthesis. To create the dual promoter vector (pDual) for siRNA expression, the promoter regions of the mouse small nuclear RNA U6 and the human H1 RNA (the RNA component of RNase P1) were amplified by PCR and cloned into pBluescript SK vector in opposite directions as shown in Fig. 1A. A HindIII site upstream of the U6 RNA transcription start site and a BglII site upstream of the H1 RNA transcription start site were created by site-directed mutagenesis (Stratagene). To construct the gene-specific siRNA expression plasmids, a pair of 35- to 37-base oligonucleotides were annealed and ligated into pDual digested with BglII and HindIII (Fig. 1 A). These oligonucleotides contain 19- to 21-nt gene-specific sequences flanked by five As on the 5′ side and five Ts on the 3′ side, as well as the restriction sites, HindIII and BglII, respectively. The sequences for the siRNA-encoding oligonucleotides are provided in Supporting Text, which is published as supporting information on the PNAS web site.

Fig. 1.

pDual siRNA expression system. (A) Strategy for generating siRNA by using two opposing Pol III promoters. To construct the pDual vector, the mouse U6 and human H1 promoter sequences were cloned into pBluescript SK in opposite directions. Appropriate mutations were made to define termination signals for siRNA transcription or facilitate inserting siRNA-encoding sequences. To create a gene-specific siRNA expression plasmid, a pair of complementary oligonucleotides (35–37 nt) were annealed and ligated into pDual digested with BglII and HindIII. (B) A single-step PCR strategy for producing siRNA expression cassettes based on the pDual system. An oligonucleotide encoding the desired siRNA sequence was used to bridge the U6 and H1 promoter fragments (templates). In addition, two 31-mer universal primers (common to all PCRs) complementary to the 5′ ends of the U6 and H1 promoters were also added to the PCR. Two separate cycling steps with different annealing temperatures were used (see Materials and Methods).

Transfection and Gene Silencing Reporter Assays. P19 mouse embryonic carcinoma cells (American Type Tissue Culture Collection, CRL-1825) were cultured in MEM-α (GIBCO/BRL) supplemented with 10% FBS. HEK293T and HeLa cells were cultured in DMEM supplemented with 10% FBS. P19 and HEK293T cells were transfected with FuGENE 6 (Roche Molecular Biochemicals). HeLa cells were transfected with Lipofectamine 2000 (Invitrogen) as directed by the manufacturer. For gene silencing experiments with transfected firefly Luc, pGL3 and pRL-SV40 were cotransfected with the siRNA expression plasmids or PCR fragments at a ratio of 2:1:20. Cells were lysed 48 h after transfection. Luc activity was measured with the Dual-Luciferase Reporter assay system (Promega). Renilla Luc activity was used as a control for normalization.

Immunohistochemistry. Cells were cultured and transfected in 96-well plates; 3–5 days after transfection the cells were fixed with 4% paraformaldehyde in PBS for 20 min at room temperature. The cells were washed with PBS three times, and then stained overnight at 4°C with primary antibody diluted with PBS containing 0.3% Triton X-100 and 5% horse serum. Monoclonal mouse antilamin A/C antibody (Santa Cruz Biotechnology) was used at a dilution of 1:300. Monoclonal mouse antibody against neuronal class III β-tubulin (Covance, Princeton) was used at a dilution of 1:500. After washing with PBS containing 0.1% Triton X-100, cells were incubated for 2 h with the secondary antibody, Cy3-Donkey anti-mouse IgG (Jackson Immuno-Research). Cells were then washed three times with PBS. Images were analyzed by fluorescence microscopy (Nikon eclipse TE2000-U) and photographed with a digital camera.

PCR for Production of siRNA Expression Cassettes. The U6 and H1 promoter sequences were amplified by PCR from the siRNA expression plasmid by using the following primers: H1 forward-GTAATACGACTCACTATGCGAACGCTG-ACGTCATCAAC; H1 reverse-TTTTTAGATCTGTCTCATACAG; U6 forward-GGAATCAGCTATGACCATGTTACGATCCGACGCCGCCATCTC; U6 reverse-CTTTTTAAGCTTTTCTCCAAGG. To produce siRNA expression cassettes (Fig. 1B), the PCRs were performed with the H1 and U6 promoters as templates, two universal primers and one gene-specific primer. The two universal primers are common to all PCRs: the universal forward primer is complementary to the 5′ end of the U6 promoter (GGAATCAGCTATGACCATGTTACGATCCG), and the universal reverse primer is complementary to the 5′ end of the H1 promoter (GTAATACGACTCACTATGCGAACGCTGACG). The gene-specific primer is unique for each siRNA expression cassette: the 5′ region of this oligonucleotide contains a 22-nt sequence complementary to the H1 promoter sequence (CTGTATGAGACAGATCTAAAAA). This sequence is followed by a 19-nt siRNA encoding sequence, and then a 20-nt sequence complementary to the U6 promoter sequence (TTTTTAAGCTTTTCTCCAAG). All of the PCRs were carried out as follows: 3 min at 94°C; 1 min at 94°C, 1 min at 55°C, and 1 min at 72°C for 5 cycles; followed by 1 min at 94°C, 1 min at 63°C, and 1 min at 72°C for 36 cycles. The resulting PCR products are ≈650 bp and >95% purity as judged by agarose gel electrophoresis. The PCR products were then purified with a QIAquick PCR purification kit (Qiagen, Chatsworth, CA).

Construction of siRNA Expression Cassette Library and Screening for Regulators of NF-κB Transcription Activity. About 8,000 genes were selected as initial targets for the construction of a large arrayed siRNA expression cassette library. Two siRNA-encoding sequences were selected for each gene. The same flanking sequences used in the PCR protocol were added to each side of the siRNA-encoding sequences. These siRNA-specific primers were synthesized by Integrated DNA Technologies (Coralville, IA). PCRs were carried out in 96-well plates by using the same conditions as described above. The PCR products were purified (QIAquick96), normalized, and spotted into 384-well plates for cell-based phenotypic screens. To screen for regulators of NF-κB transcriptional activation by tumor necrosis factor α (TNF-α), the pNF-κB-Luc reporter plasmid (Clontech) was cotransfected with siRNA expression cassettes into HEK293T cells. Seventy-two hours after transfection, cells were stimulated with 10 ng/ml TNF-α (Sigma) for 12–14 h, and Luc activity was subsequently measured with the Bright-Glo Luciferase Assay Kit (Promega). To screen for activators of NF-κB basal transcriptional activity, cells were left unstimulated, and Luc activity was measured 96 h after transfection.

Results and Discussion

Design of Dual Promoter Vector. The design of the siRNA expression vector (pDual) features two opposing RNA pol III promoters to direct the transcription of the sense and antisense siRNA sequences from an interposed gene-specific sequence. To avoid vector instability that might be caused by two complementary promoter sequences flanking the siRNA-encoding sequence, two different pol III promoters, the mouse U6 and human H1 promoters, were chosen. Five Ts were inserted at the 3′ end of the two opposing promoters as a termination signal, because it is known that for the H1 promoter the last five nucleotides near the transcriptional start site can be substituted by unrelated sequences without affecting promoter activity (19, 20). To construct the pDual vector, the mouse U6 and human H1 promoter sequences were cloned into pBluescript SK in opposite directions (Fig. 1A). To facilitate the cloning of siRNA-encoding sequences into the vector, HindIII and BglII sites were created in the promoter sequences flanking the insertion site by site-directed mutagenesis. All mutations created in the promoter sequences, including the –TTTTT–substitution, did not affect the position of the transcriptional start site relative to the TATA box. To create a gene-specific siRNA expression plasmid, a pair of complementary oligonucleotides (35–37 nt) were annealed and ligated into pDual digested with BglII and HindIII. In the resulting plasmid, 19–21 bp of the siRNA-encoding sequence are flanked by five As at the 5′ end and five Ts at the 3′ end. Once transfected into mammalian cells, the sense and antisense strands are transcribed by two opposing promoters on the same template, resulting in 19- to 21-bp RNA duplexes with a uridine overhang at the 3′ end, closely resembling the digested product of dicer.

PCR Protocol for High-Throughput Production of siRNA Expression Cassettes. To produce the siRNA expression cassettes, a single-step PCR protocol was developed based on the dual promoter vector (Fig. 1B). An oligonucleotide encoding the desired siRNA sequence was used to bridge the U6 and H1 promoter fragments. In addition, two 31-mer universal primers complementary to the 5′ ends of the U6 and H1 promoters, respectively, were added to the PCR. The latter two primers are common to all PCRs. To efficiently produce the desired PCR product, a PCR protocol was used that involves two separate cycling steps with different annealing temperatures. In the first five cycles, the three primers were annealed at 55°C with the corresponding templates. In the following 36 cycles, the annealing temperature was raised to 63°C such that only the two universal primers were able to anneal with their templates, resulting in amplification of only the full-length siRNA expression cassette. Under these conditions, the major PCR product is the 650-bp, full-length siRNA expression cassette (>95% purity as judged by agarose gel and sequencing).

Efficient Inhibition of both Transfected and Endogenous Gene Expression. To determine whether the pDual system can efficiently produce functional siRNAs, a reporter assay for firefly Luc was used. siRNA sequences 19 nt in length that correspond to three different regions of the firefly Luc gene were cloned into pDual. For comparison, one of the target sequences was also used to prepare a short hairpin siRNA by using the pSUPER vector. These siRNA expression vectors were cotransfected with firefly Luc and Renilla Luc expression vectors into 293T cells at a ratio of 20:2:1. Cells were lysed 48 h after transfection, and firefly Luc and Renilla Luc activity were measured with the Dual-Luciferase Reporter assay system (Promega). As shown in Fig. 2, siRNAs transcribed by pDual system encoded by sequences 1, 2, and 3 (see Materials and Methods) can efficiently suppress transfected Luc gene expression with a 70–90% reduction in Luc activity in comparison to the empty pDual control (4). These results are similar to those produced by the hairpin siRNA expressed by the single H1 promoter construct (5). Renilla Luc activity was not affected by the expression of siRNAs that target the firefly Luc coding sequence, demonstrating that the effect is gene specific.

Fig. 2.

Specific and efficient suppression of transfected firefly Luc by siRNAs expressed by using the pDual system in 293T cells. siRNA sequences of 19 nt in length that correspond to three different regions (lanes 1–3) of the firefly Luc gene were chosen as the target sequences and cloned into pDual; empty pDual was used as control (lane 4); and short hairpin RNA was expressed by pSuper (lane 5). For comparison, siRNA coding sequences in lanes 1 and 5 are the same.

We next determined whether siRNAs expressed by the pDual system can efficiently inhibit endogenous gene activity. It has been shown that lamin A/C expression can be efficiently suppressed by synthetic siRNAs or hairpin siRNAs transcribed by a single U6 promoter (21, 22). To test the pDual system, a pair of oligonucleotides that contain 21-nt siRNA coding sequences targeting lamin A/C were inserted into pDual to generate the pDual-lamin A/C. HeLa cells were then transfected with either pDual-lamin A/C or the empty pDual vector as a control. To identify transfected cells, cells were cotransfected with pCMV-GFP. Lamin A/C expression levels were significantly reduced (≥70%) in cells transfected with pDual-lamin A/C (indicated by GFP) (Fig. 3). In some cells, lamin A/C expression was completely abolished (Fig. 3B). This finding is in striking contrast to cells transfected with pDual alone (Fig. 3E), where no detectable change in lamin A/C expression was found. We have also shown that siRNA expressed with dual promoter vector can efficiently suppress βIII-tubulin expression and block the neuronal differentiation induced by overexpression of a transcription factor MASH1 (23) (see Supporting Text and Fig. 5, which is published as supporting information on the PNAS web site). These results again demonstrate that siRNAs transcribed in vivo by two opposing RNA pol III promoters in the pDual system can specifically and efficiently suppress endogenous gene expression. The dual promoter vector can also be used to express long dsRNA in cells (see Supporting Text and Fig. 6, which is published as supporting information of the PNAS web site).

Fig. 3.

Efficient suppression of lamin A/C by transfected pDual-lamin A/Cin HeLa cells. (A–C)Efficient suppression of lamin A/C in HeLa cells cotransfected with pDual-lamin A/C and pCMV-GFP. (D–F) No suppression of lamin A/C in HeLa cells transfected with pDual and pCMV-GFP. (A and D) GFP expression identifies the transfected cells. (B and E) Lamin A/C was stained with monoclonal antilamin A/C antibody. (C and F) The merged images of GFP and lamin A/C staining. (Magnifications: ×200.)

Gene Suppression by the PCR-Derived siRNA Expression Cassettes. The suppression efficiency of the PCR-derived siRNA expression cassettes derived from the pDual system was also assayed by using the Luc reporter. Either the PCR product or pDual-luc vector was contransfected into HEK293T or P19 cells with pGL3 and pRL-SV40 in a ratio of 10:1:0.5. As shown in Fig. 4 A and B, the PCR product inhibits the expression of the transiently transfected Luc gene by ≈70–80% (Fig. 4 A and B, lane 4), comparable to suppression in cells transfected with pDual-luc (Fig. 4 A and B, lane 2). In these experiments, the same amounts of PCR product and pDual-luc were used. Even though the molar ratio of PCR product to the reporter gene is much higher than the ratio of pDual-luc to the reporter, the suppression efficiency of the PCR product is slightly lower than that of vector-based approach. This could result from lower transfection efficiency, lower transcriptional efficiency, and/or instability of the PCR product in cells.

Fig. 4.

(A and B) Specific and efficient suppression of firefly Luc expression by PCR-derived siRNA expression cassettes based on the pDual system. Comparison of gene suppression effects by siRNAs derived from pDual vector and PCR fragments in HEK293T cells (A) or P19 cells (B). Cells were transfected with pGL3/pRL-SV40 and pDual-Luc (lane 2), or pGL3/pRL-SV40 and PCR-derived siRNA expression cassette (lane 4). pDual and PCR fragment derived from pDual are used as control (lanes 1 and 3). (C–H)Efficient suppression of lamin A/C by PCR-derived siRNA expression cassettes transfected in HeLa cells. (C–E) HeLa cells were cotransfected with PCR-derived lamin A/C siRNA expression cassettes and pCMV-GFP. (F–H) PCR-derived firefly Luc siRNA expression cassettes and pCMV-GFP. (C and F) GFP marks the transfected cells. (D and G) Lamin A/C was stained with monoclonal antilamin A/C antibody. (E and H) Merged images of GFP and lamin A/C staining. (Magnifications: ×200.)

We next determined whether the PCR product can suppress endogenous gene activity. The PCR-derived lamin A/C siRNA expression cassettes were again cotransfected with pCMV-GFP into the HeLa cells and the Luc siRNA expression cassette was used as a control. Three days after transfection, cells were fixed and immunostained with antilamin A/C antibody. As shown in Fig. 4 C–H, cells transfected with lamin A/C siRNA expression cassettes have significantly reduced lamin A/C expression (≥70%) (Fig. 4D), whereas the control Luc siRNA expression cassette had no effect on lamin A/C expression (Fig. 4G). These results demonstrate that the PCR-derived siRNA cassettes based on our pDual system are sufficiently stable in cells to temporally suppress endogenous gene expression.

Screening for Regulators of NF-κB Signaling Using a siRNA Expression Cassette Library. The PCR-derived siRNA expression cassette provides a high-throughput method for generating large libraries of gene-specific siRNAs for genomewide loss-of-function cellular screens. Thus, we set out to construct a large siRNA expression cassette library targeting ≈8,000 human genes from the public UniGene library with two targeting sequences per gene. These genes represent a large fraction of the proteins known to be involved in biosynthesis, metabolism, signal transduction, gene regulation, and cell cycle control; the distribution of these genes is shown in Scheme 1 (a list of targeted genes can be found at our web site, schultz.scripps.edu/siRNA/genelist.html). The target sequences for each gene were chosen based on the guidelines described (4, 24). Coding regions were searched for the 23-nt sequence motif AAG(N17)(C/A/G)TT (N, any nucleotide), and sequences were selected with ≈50% G/C content (40–65% is allowed range). If no suitable sequences were found, no constraints were placed on the first and last two nucleotides of the 23-nt sequence. However, no stretch of four or more of the same nucleotides was allowed in any sequence. All sequences were blasted against the same UniGene library, and any sequence with >89% homology to any other gene in the middle N19 nucleotides was excluded. The same procedure was carried out for AA(T/C)N17(C)TT in the antisense sequence.

Scheme 1.

Gene distribution of siRNA expression cassette library.

To demonstrate the utility of this siRNA expression cassette library as a tool to annotate gene function, the arrayed library was initially screened for genes involved in the NF-κB signaling pathway in mammalian cells. The transcription factor NF-κB controls many diverse cellular processes, including growth, development, inflammation, immune response, apoptosis, and oncogenesis (25, 26). To screen for regulators of NF-κB transcriptional activation by TNF-α, the pNF-κB-Luc reporter plasmid was cotransfected with siRNA expression cassettes into HEK293T cells. Seventy-two hours after transfection, cells were stimulated with 10 ng/ml TNF-α (Sigma) for 12–14 h, and Luc activity was subsequently measured. To screen for activators of NF-κB basal transcriptional activity, cells were left unstimulated, and Luc activity was measured 96 h after transfection.

Ninety-four genes were initially identified from these two screens that differentially modulate the reporter activity. Of these, 20 genes (Table 1), which when suppressed result in a significant increase (4- to 6-fold) in basal NF-κB activity, or a significant increase (≥6 fold) or reduction (≥70%) in TNF-α-stimulated activity, were initially chosen for further characterization. To confirm that these are not off-target effects, additional siRNA sequences from different regions of each gene were designed and tested under the same conditions. Expression of 17 of these 20 genes could be suppressed by three or more distinct sequences (Table 2 and Fig. 7, which are published as supporting information on the PNAS web site). Eight of the genes identified encode proteins that are known to be involved in the NF-κB signaling pathway, including NF-κB1, RelA, IKKβ, TNF-α receptor, the F-box protein for IκB (β-TrCP), phosphatase 2A, PKC delta, and IL-1 receptor-associated protein (IL-1RAP) (25–28).

Table 1. Genes used for characterization.

Gene name	GenBank accession no.	RNAi effect
Known components of NF-κB signaling pathway
NF-κB1	NM_003998	Inhibition of response to TNF
Rel A	NM_021975	Inhibition of response to TNF
IKKβ	AF080158	Inhibition of response to TNF
TNFRSF1A	NM_001065	Inhibition of response to TNF
β-TrCP	NM_033637.	Inhibition of response to TNF
PPP2R1A	NM_014225	Increase basal activity/synergistic activation
PKC δ	NM_006254	Increase basal activity/synergistic activation
IL1RAP	NM_002182	Increase basal activity/synergistic activation
Genes with previously unreported function in NF-κB signaling
BCL-2L13	NM_015367	Increase basal activity
DAPK1	NM_004938	Increase basal activity
TBC1D5	NM_014744	Increase basal activity
Rab11-FIP3	NM_014700	Increase basal activity/synergistic activation
CDC25B	NM_004358	Inhibition of response to TNF
FAP48	NM_007070	Increase basal activity
Son A	NM_003103	Increase basal activity/synergistic activation
STAM2	NM_005843	Increase basal activity
CARP	NM_014391	Increase basal activity/synergistic activation
DUSP5	NM_004419	Increase basal activity/synergistic activation
EIAVL1	NM_001419	Increase basal activity/synergistic activation
APBA2	NM_005503	Increase basal activity/synergistic activation

Of the remaining genes some may be involved in cellular processes that indirectly affect the NF-κB pathway. For example, two apoptosis inducers, BCL-2-like protein 13 and death-associated protein kinase (DAPK2), when suppressed by RNAi may enhance cell proliferation or related processes, resulting in basal NF-κB transcriptional activation. Two small GTPase Rab-related proteins, TBC1D5 (a possible Rab-GAP) and Rab-interacting protein (Rab11-FIP), appear to be positive regulators of NF-κB pathway and may affect intracellular protein trafficking of members of the pathway. And finally, ELAV1, a negative regulator of the NF-κB pathway, may function by controlling the mRNA stability of some components in the pathway.

Several genes identified in the screen may have previously unrecognized roles in the NF-κB signaling, including FAP48, a FK506-binding protein (FKBP)-associated protein (29, 30). FKBP binds and inhibits calcineurin A, a protein phosphatase that can augment the basal transcriptional activity of NF-κB by dephosphorylating IκBβ in response to mitchondria stress. The interaction of FAP48 with FKBP may be required for formation of the FKBP-calcineurin complex; suppression of FAP48 may release calcineurin from FKBP, and thereby increase the basal transcriptional activity of NF-κB. Alternatively, because FAP48 is also an antiproliferative molecule, suppression of FAP48 could also increase cell proliferation and lead to enhanced activity. The transcriptional regulator Son A was identified in the screen as a negative regulator of the NF-κB pathway. Son A binds to the regulatory elements of human hepatitis B virus (HBV) and represses the transcription of HBV genes and production of HBV viron (31). It also protects cells from apoptosis induced by staurosporine or Bax overexpression (32). Suppression of Son A by RNAi strongly enhances the NF-κB basal transcription activity, possibly indicating a role in regulating transcription of some components in the pathway. Finally, DUSP 5 is a dual-specificity phosphatase, whose expression is regulated by p53. Overexpression of DUSP 5 suppresses the growth of several types of human cancer cells by dephosphorylation of proteins involved in signal transduction (33) and may act on members of NF-κB pathways. Whether these genes and also other genes on the list are directly involved in the NF-κB signaling or play broader roles in the cellular process will require further investigation.

Conclusion

We have designed and constructed a dual promoter system (pDual) for expression of siRNAs in mammalian cells that can specifically and efficiently suppress gene functions. We have developed a single-step PCR protocol to produce siRNA expression cassettes in a high-throughput fashion and shown that these siRNA expression cassettes can also specifically and efficiently suppress the expression of both transfected and endogenous genes. This single-step PCR approach is efficient and cost-effective and makes high-throughput production of siRNA expression cassette libraries for genomewide functional gene annotation practical. A high-throughput cell-based screen of an arrayed siRNA expression cassette library identified known components of the NF-κB signaling pathways as well as genes that may have previously unrecognized roles in the regulation the NF-κB transcription activity. However, other components of the NF-κB signaling pathway were not found. This may reflect a need for developing better algorithms for choosing siRNA-encoding sequences. Alternatively, the siRNA expression cassette library contains only two siRNA-encoding sequences per gene, which may limit the success rate in suppressing the expression of any given gene. Finally, the PCR protocol described here for generating siRNA expression cassettes also allows the facile addition of restriction sites or Gateway sequences to the ends of PCR fragments, which can be used to subsequently clone these PCR fragments into vectors to generate a virus-based siRNA expression library.