![[Feedback]](/icons/banner/feedbackACT.gif){kind=icon}
![[For Subscribers]](/icons/banner/subscriberACT.gif){kind=icon}
![[Archive]](/icons/banner/archiveACT.gif){kind=icon}
![[Search]](/icons/banner/searchACT.gif){kind=icon}
![[Contents]](/icons/banner/tocACT.gif){kind=icon}
|
|
|
|
|
||
Vol. 17, Issue 11, 4837-4845, November 2006
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
*Department of Biochemistry and Molecular Genetics, University of Virginia School of Medicine, Charlottesville, VA 22908; and Department of Computer Sciences, University of Virginia School of Engineering and Applied Science, Charlottesville, VA 22904
Submitted April 24, 2006;
Accepted August 28, 2006
Monitoring Editor: Charles Boone
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
). Specific gene silencing can be triggered by soaking worms or Drosophila cells in a solution containing RNA duplexes of target mRNAs (Tabara et al., 1998; Clemens et al., 2000). RNAi was used to perform large-scale screens in these organisms (Fraser et al., 2000; Gonczy et al., 2000; Maeda et al., 2001; Kamath et al., 2003; Bettencourt-Dias et al., 2004; Boutros et al., 2004). These screens have identified genes involved in cell division, apoptosis, and fat metabolism. In mammalian cells, however, introduction of long double-stranded RNA (>30 nucleotides) triggers the antiviral interferon response and subsequent cell death (Stark et al., 1998). This problem was solved by the discovery that duplexes of 21-nucleotide RNA termed siRNAs (small interfering RNAs) can trigger RNAi pathway in mammalian cells without activating the antiviral response (Elbashir et al., 2001). siRNA can be generated by chemical synthesis or enzymatic digestion of target mRNA duplexes in vitro (Yang et al., 2002) or by expressing short hairpin RNAs in mammalian cells (Brummelkamp et al., 2002; Sui et al., 2002).
The human genome sequencing project revealed that human cells contain 
20,000–25,000 protein-coding genes (International Human Genome Sequencing Consortium, 2004) and has encouraged genome-wide functional genomics in mammals. Several groups have constructed RNAi libraries to perform loss-of-function genetics in cultured mammalian cells (Berns et al., 2004; Kittler et al., 2004; Paddison et al., 2004; Moffat et al., 2006). Libraries of short hairpin RNAs or endoribonuclease-prepared short interfering RNAs (siRNAs) were used in selection and screening assays, respectively, to identify genes important for cancer cell phenotypes (Kolfschoten et al., 2005; MacKeigan et al., 2005; Westbrook et al., 2005). Although the response to silencing of a few genes is different between two Drosophila cell lines, suggesting that cell lines from different backgrounds respond differently (Kiger et al., 2003), no attention has been paid to exploiting the vast heterogeneity of mammalian cell lines in siRNA screens. We reasoned that if phenotypes of multiple cell lines after RNAi-mediated gene silencing can be compared in a quantitative and high throughput manner, we can identify genes with differential requirements between cell lines. Targeted comparative RNAi (TARCOR) reported here can be used to perform such comparative functional genomics in multiple human cell lines. The comparison required the development of criteria for quantitative reproducibility of results. In addition, instead of a global analysis on a heterogeneous collection of genes, TARCOR targets a narrower set of genes relevant to the phenotype being studied. For example, the genes can be selected from previous studies such as microarray-based gene expression analysis. We demonstrate that TARCOR can be used to identify genes that are differentially required for proliferation of two human cell lines. Furthermore, the differential sensitivity to several of the genes could be due to the activation of p53 in one cell line and not the other. For example, inhibition of ribosome biogenesis appears to cause a G1/S arrest by a p53-dependent pathway in the p53+ MCF10A cell. Besides emphasizing the importance of p53 in a cell's response to depletion of many genes, the results highlight how such targeted comparative screens can find biologically relevant differences between cells and conversely, how the heterogeneity of mammalian cell lines can be exploited to add value to siRNA screens.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
|
-actin (loading control). In vitro kinase assays were performed as reported previously (Machida et al., 2005). Anti-cyclin E (HE111; Santa Cruz) was used for immunoprecipitation.
Bromodeoxyuridine ELISA
Cells were incubated with 10 µM bromodeoxyuridine(BrdU) for 15 min to 1 h. Cells were washed once with PBS and fixed with FixDenat (Roche, Indianapolis, IN) for 30 min. After washing once with PBS, the plates were blocked with 3% BSA in PBS for 1 h. The plates were then incubated with HRP-coupled anti-BrdU antibody (Roche) diluted in 3% BSA in PBS for 1 h. After washing three times with PBS containing 0.1% TX-100, the plates were incubated with TMB substrate (Pierce, Rockford, IL) for 5–10 min. The reactions were stopped by adding 1 M H2SO4, and the absorbance at 450 nm was measured.
Data Preprocessing and Analysis
Each assay plate contained four wells of negative control (luciferase; GL2) and two wells of positive control (ORC2) siRNAs for normalization. To normalize values of BrdU incorporation, we calculated the inhibition index (%) using the following equation: Inhibition index of gene X (%) = (GL2av – X)/(GL2av – ORC2av) x 100, where X, GL2av, and ORC2av represent BrdU incorporation (absorbance at 450 nm) of gene X and average of GL2 and ORC2, respectively. Genes with a SD of inhibition indices (from 3 technical replicates) greater than a cutoff value were eliminated to select technically reproducible data. To select biologically reproducible data, inhibition indices from two screens were plotted and the distance of each gene from the y = x line, representing ideal behavior, was calculated and plotted. Genes with distances in the top 5% or bottom 5% of the distribution were eliminated to select biologically reproducible data. The cutoff value that eliminates 10% of the worst performing genes in the biological replicates was used as the cutoff to eliminate the worst performing genes in the technical replicates. Hierarchical clustering was performed on the response of cells to the silencing of individual genes. To select genes that have reliably differential effects on MCF10A and PC3, we calculated the differential inhibition index using the following equation: Differential inhibition index of a gene = (AVM – AVP)/(SDM + SDP), where AVM and AVP represent the average of inhibition indices of a gene in MCF10A and PC3, respectively, and SDM and SDP represent the SD of inhibition indices for each cell line.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
) or E2F-regulated genes (Ren et al., 2002; No. 1-228 in Supplementary Table S1). Besides many well-characterized cell cycle–related genes, the gene-set contains many uncharacterized genes so that we have a chance to identify new genes involved in proliferation control. In addition, we included 29 uncharacterized ATPases to test if we can expand TARCOR analysis to gene sets selected by other criteria (No. 229-257 in Supplementary Table S1). Figure 1A represents the scheme of TARCOR analysis of these genes in human cell lines. siRNAs were transfected in 96-well plates and incorporation of BrdU measured after 72 h to quantitate the proliferation of the cells. Because we compare BrdU incorporation per well, we could identify genes essential for viability as well as cell proliferation. In this study, we compared a breast epithelial cell line, MCF10A, and a prostate cancer cell line, PC3.
|
Because the primary screen was performed on very small populations of cells in 96-well plates, [3H]thymidine incorporation in 24-well plates was used to confirm a subset of the results (Supplementary Figure S1A). RT-PCR confirmed the reduction of target mRNAs after RNAi of a subset of the genes (Supplementary Figure S1B). For genes that were followed up, we eliminated the possibility of off target activity of siRNAs by confirming the inhibition of cell growth by a second RNA duplex against a different sequence of the gene or by transfecting SMARTpools of siRNAs (Dharmacon; Supplementary Figure S1C and unpublished data).
Comparison of Gene Requirement in MCF10A and PC3
We could now compare the requirement of individual genes in the two cell lines. By intersecting gene sets that passed the reproducibility screen (191 and 223 genes in MCF10A and PC3, respectively), we obtained 172 genes with reproducible inhibition indices in both MCF10A and PC3. A scatter plot of the inhibition indices of the 172 targeted genes from two screens in the same cell line shows that the inhibition indices are close to a y = x diagonal, confirming that data are highly reproducible in a given cell line (correlation coefficients of 0.930 and 0.958 in MCF10A and PC3, respectively; Figure 2, A and B). The inhibition indices are more scattered when MCF10A and PC3 data were compared with each other (r = 0.630; Figure 2C). Seventy-three percent of the genes were within the cutoff lines of biological reproducibility for MCF10A (Figure 2C, genes between two dashed lines), suggesting that a large group of genes behave similarly in this assay in the cells from two different lineages. In contrast, 27% of the genes (46 genes) were outside the cutoff lines, suggesting that RNAi against a subset of genes affects the two cell lines differently. A hierarchical cluster analysis of the inhibition indices for 172 genes identified clusters of genes whose knockdown had differential effects in the two cell lines (Figure 2D and data in Supplementary Table S2).
|
|
). Because p53 is induced by a variety of cellular stresses, we wondered whether the positive p53 status of MCF10A might contribute to the selective toxicity seen upon depletion of the genes listed above. Of the five siRNAs that spared PC3 cells despite knockdown to levels below that in MCF10A, four (DDX21, EBNA1BP2, NOL5A, and S1P1) induced p53 protein more than fivefold in MCF10A (Figure 4). In addition, three more of the siRNAs that spared PC3 (MCM3, NET1, and UMPS) induced p53 protein more than fivefold. Therefore p53 is activated by 7 of the 10 genes that were selectively toxic to MCF10A.
|
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Comparative Analysis of Cell Lines
The quantitative comparison of the cellular responses to knockdown of genes in two human cell lines allowed the systematic discovery of different rate-limiting steps in cells of different lineages and different genetic backgrounds. The key to a successful comparison of the cell lines was the establishment of a highly reproducible screening method involving high throughput siRNA transfection and measurement of BrdU incorporation. Furthermore, we applied a strict quality control filter so that only highly reproducible data were used for the subsequent comparison analysis. An unexpectedly high percentage of genes (>25%) showed a differential effect on survival of the two cell lines upon depletion by RNAi. This result suggests that cells of different lineages and genetic backgrounds have different rate-limiting genes even for the fundamental phenotype of cell proliferation. Thus TARCOR is particularly suited for discovering genes that are differentially rate-limiting among multiple cell lines.
The difference in the genetic background of cells is certainly responsible for some of the differential effects of siRNAs. Knockdown of many genes that affect MCF10A (p53+) but not PC3 (p53–) induces p53 in MCF10A cells. The cell cycle arrest after EBNA1BP2 depletion was indeed dependent on p53 induction in MCF10A. Thus, p53 status of cell lines seems to be a major determinant of differential cellular response to depletion of a growth-related genes.
Genes Essential for Proliferation of Human Cells
Most of the current screens for essential genes have been performed in model organisms such as yeasts, flies, or C. elegans, where the genetic homogeneity has led to the assumption that the same genes will be uncovered as essential. The genetic and epigenetic variability of mammalian cells, however, must be taken into account when devising screens for essential genes in human cells. An unexpected benefit of TARCOR on mammalian cells was a significant expansion of the number of hit genes when the analysis was expanded to two cell lines. We obtained different hit genes in the two cell lines, with the total pool of hit genes increasing by 45% (116 vs. 80 genes) by pooling results from two cell lines. Thus it is important to use multiple cell lines when screening genes in human cell lines. The list of hit genes included many expected genes involved in the cell cycle and in DNA, RNA, and protein metabolism. Thus the 15 genes for which no function has been ascribed and which were identified as essential for proliferation in this study have the potential to be involved in these critical processes.
SiRNA screens are hypomorphic genetic screens and not null screens. In such screens a gene could be deemed essential for proliferation, either because the lower level of the gene product is insufficient for an essential function in cell growth or because the lower level triggers cellular checkpoint mechanisms that arrest the cell cycle or induce apoptosis. One way to distinguish the two mechanisms is to test if impairment of a checkpoint pathway can restore cell proliferation. In this article, EBNA1BP2 is an example of a gene that scores as essential for proliferation because of the second type of mechanism. Codepletion of p53 by RNAi can rescue the cell cycle defect of EBNA1BP2-depleted cells, indicating that the decrease in the EBNA1BP2 is not sufficient to be the direct cause of the cell cycle arrest in MCF10A cells.
How does EBNA1BP2 depletion activate the p53 pathway? It has been recognized that any stress on rRNA production arrests the cell cycle through a p53-mediated pathway (Pestov et al., 2001). Because the yeast homolog of EBNA1BP2 is involved in rRNA processing (Huber et al., 2000; Tsujii et al., 2000), the p53 induction could be in response to stress on rRNA production. Another possibility is suggested by the fact that EBNA1BP2 binds to the Epstein-Barr virus nuclear antigen 1 (EBNA1) protein, which is known to destabilize p53 by inhibiting the latter's interaction with HAUSP/USP7, a deubiquitination enzyme for p53 (Saridakis et al., 2005). Although the cells used in our studies do not contain EBNA1, an intriguing possibility is that EBNA1BP2 (or a cellular interaction partner) is required to destabilize p53 by similar mechanisms.
Targeted Screen
DNA microarrays have successfully identified large sets of differentially expressed genes. Focusing on cell proliferation-related genes previously identified by microarray studies allowed us to obtain a significantly higher hit rate in our analysis, suggesting that the high throughput targeted RNAi analysis is particularly useful for following up on the accumulating microarray data. Hundreds of genes are differentially expressed in specific cancer cells in microarray studies, but there is no easy method for sorting through these genes to identify those relevant to cancer cell phenotypes. Our results suggest that the screens targeted on genes differentially expressed in cancer cells might be an efficient way of finding those that are critical for cancer cell proliferation.
TARCOR in Drug Discovery
The key issue for cancer drug development is selectivity for the target cell types. One way to add selectivity to drugs is to choose a target protein that is selectively rate-limiting for the target cells. Thus TARCOR can be an effective screening method for discovery of cancer-specific drug targets among genes that are differentially expressed in cancer cells in microarray studies. Another application will be the comparison of two cell lines that are isogenic except for a single cancer-causing mutation. The potential for such a screen is revealed by our discovery that inhibition of ribosome biogenesis is selectively inhibitory to p53+ but not p53– cells. TARCOR screens for genes that are synthetically lethal with cancer-related loss-of-function mutations can be executed, for example, by comparison of cells with functional and nonfunctional BRCA1 (Scully et al., 1999) or of cells that are p53+/+ and p53–/– (Bunz et al., 1998). Such screens are expected to identify drug targets that are specific to cancer cells with those mutations.
So far, drugs are mostly small chemicals that bind to receptors or enzymes. However, siRNAs themselves might be used directly as drugs in future (Soutschek et al., 2004; Zimmermann et al., 2006). In that case TARCOR will become even more useful because the differentially inhibitory siRNA is immediately viable as a drug.
|
ACKNOWLEDGMENTS |
|---|
|
Footnotes |
|---|
The online version of this contains supplemental material at MBC Online (http://www.molbiolcell.org). 
This article was published online ahead of print in MBC in Press (http://www.molbiolcell.org/cgi/doi/10.1091/mbc.E06-04-0340) on September 6, 2006.
Address correspondence to: Anindya Dutta (ad8q{at}virginia.edu)
Abbreviations used: RNAi, RNA interference; siRNA, small interfering RNA; TARCOR, targeted comparative RNAi
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Bettencourt-Dias, M., et al. (2004). Genome-wide survey of protein kinases required for cell cycle progression. Nature 432, 980–987.[CrossRef][Medline]
Boutros, M., Kiger, A. A., Armknecht, S., Kerr, K., Hild, M., Koch, B., Haas, S. A., Consortium, H. F., Paro, R., Perrimon, N. (2004). Genome-wide RNAi analysis of growth and viability in Drosophila cells. Science 303, 832–835.
Brummelkamp, T. R., Bernards, R., Agami, R. (2002). A system for stable expression of short interfering RNAs in mammalian cells. Science 296, 550–553.
Bunz, F., Dutriaux, A., Lengauer, C., Waldman, T., Zhou, S., Brown, J. P., Sedivy, J. M., Kinzler, K. W., Vogelstein, B. (1998). Requirement for p53 and p21 to sustain G2 arrest after DNA damage. Science 282, 1497–1501.
Carroll, A. G., Voeller, H. J., Sugars, L., Gelmann, E. P. (1993). p53 oncogene mutations in three human prostate cancer cell lines. Prostate 23, 123–134.[Medline]
Clemens, J. C., Worby, C. A., Simonson-Leff, N., Muda, M., Maehama, T., Hemmings, B. A., Dixon, J. E. (2000). Use of double-stranded RNA interference in Drosophila cell lines to dissect signal transduction pathways. Proc. Natl. Acad. Sci. USA 97, 6499–6503.
Elbashir, S. M., Harborth, J., Lendeckel, W., Yalcin, A., Weber, K., Tuschl, T. (2001). Duplexes of 21-nucleotide RNAs mediate RNA interference in cultured mammalian cells. Nature 411, 494–498.[CrossRef][Medline]
Fire, A., Xu, S., Montgomery, M. K., Kostas, S. A., Driver, S. E., Mello, C. C. (1998). Potent and specific genetic interference by double-stranded RNA in Caenorhabditis elegans. Nature 391, 806–811.[CrossRef][Medline]
Fraser, A. G., Kamath, R. S., Zipperlen, P., Martinez-Campos, M., Sohrmann, M., Ahringer, J. (2000). Functional genomic analysis of C. elegans chromosome I by systematic RNA interference. Nature 408, 325–330.[CrossRef][Medline]
Gonczy, P., et al. (2000). Functional genomic analysis of cell division in C. elegans using RNAi of genes on chromosome III. Nature 408, 331–336.[CrossRef][Medline]
Huber, M. D., Dworet, J. H., Shire, K., Frappier, L., McAlear, M. A. (2000). The budding yeast homolog of the human EBNA1-binding protein 2 (Ebp2p) is an essential nucleolar protein required for pre-rRNA processing. J. Biol. Chem 275, 28764–28773.
Iyer, V. R., et al. (1999). The transcriptional program in the response of human fibroblasts to serum. Science 283, 83–87.
Kamath, R. S., et al. (2003). Systematic functional analysis of the Caenorhabditis elegans genome using RNAi. Nature 421, 231–237.[CrossRef][Medline]
Kiger, A. A., Baum, B., Jones, S., Jones, M. R., Coulson, A., Echeverri, C., Perrimon, N. (2003). A functional genomic analysis of cell morphology using RNA interference. J. Biol 2, 27.1–27.15.
Kittler, R., et al. (2004). An endoribonuclease-prepared siRNA screen in human cells identifies genes essential for cell division. Nature 432, 1036–1040.[CrossRef][Medline]
Kolfschoten, I. G., van Leeuwen, B., Berns, K., Mullenders, J., Beijersbergen, R. L., Bernards, R., Voorhoeve, P. M., Agami, R. (2005). A genetic screen identifies PITX1 as a suppressor of RAS activity and tumorigenicity. Cell 121, 849–858.[CrossRef][Medline]
Machida, Y. J., Teer, J. K., Dutta, A. (2005). Acute reduction of an origin recognition complex (ORC) subunit in human cells reveals a requirement of ORC for Cdk2 activation. J. Biol. Chem 280, 27624–27630.
MacKeigan, J. P., Murphy, L. O., Blenis, J. (2005). Sensitized RNAi screen of human kinases and phosphatases identifies new regulators of apoptosis and chemoresistance. Nat. Cell Biol 7, 591–600.[CrossRef][Medline]
Maeda, I., Kohara, Y., Yamamoto, M., Sugimoto, A. (2001). Large-scale analysis of gene function in Caenorhabditis elegans by high-throughput RNAi. Curr. Biol 11, 171–176.[CrossRef][Medline]
Moffat, J., et al. (2006). A lentiviral RNAi library for human and mouse genes applied to an arrayed viral high-content screen. Cell 124, 1283–1298.[CrossRef][Medline]
Paddison, P. J., et al. (2004). A resource for large-scale RNA-interference-based screens in mammals. Nature 428, 427–431.[CrossRef][Medline]
Pestov, D. G., Strezoska, Z., Lau, L. F. (2001). Evidence of p53-dependent cross-talk between ribosome biogenesis and the cell cycle: effects of nucleolar protein Bop1 on G(1)/S transition. Mol. Cell. Biol 21, 4246–4255.
Ren, B., Cam, H., Takahashi, Y., Volkert, T., Terragni, J., Young, R. A., Dynlacht, B. D. (2002). E2F integrates cell cycle progression with DNA repair, replication, and G(2)/M checkpoints. Genes Dev 16, 245–256.
Saridakis, V., et al. (2005). Structure of the p53 binding domain of HAUSP/USP7 bound to Epstein-Barr nuclear antigen 1 implications for EBV-mediated immortalization. Mol. Cell 18, 25–36.[CrossRef][Medline]
Scully, R., Ganesan, S., Vlasakova, K., Chen, J., Socolovsky, M., Livingston, D. M. (1999). Genetic analysis of BRCA1 function in a defined tumor cell line. Mol. Cell 4, 1093–1099.[CrossRef][Medline]
Soutschek, J., et al. (2004). Therapeutic silencing of an endogenous gene by systemic administration of modified siRNAs. Nature 432, 173–178.[CrossRef][Medline]
Stark, G. R., Kerr, I. M., Williams, B. R., Silverman, R. H., Schreiber, R. D. (1998). How cells respond to interferons. Annu. Rev. Biochem 67, 227–264.[CrossRef][Medline]
Sui, G., Soohoo, C., el Affar, B., Gay, F., Shi, Y., Forrester, W. C. (2002). A DNA vector-based RNAi technology to suppress gene expression in mammalian cells. Proc. Natl. Acad. Sci. USA 99, 5515–5520.
Tabara, H., Grishok, A., Mello, C. C. (1998). RNAi in C. elegans: soaking in the genome sequence. Science 282, 430–431.
Tsujii, R., Miyoshi, K., Tsuno, A., Matsui, Y., Toh-e, A., Miyakawa, T., Mizuta, K. (2000). Ebp2p, yeast homologue of a human protein that interacts with Epstein-Barr virus nuclear antigen 1, is required for pre-rRNA processing and ribosomal subunit assembly. Genes Cells 5, 543–553.[Abstract]
Westbrook, T. F., et al. (2005). A genetic screen for candidate tumor suppressors identifies REST. Cell 121, 837–848.[CrossRef][Medline]
Yang, D., Buchholz, F., Huang, Z., Goga, A., Chen, C. Y., Brodsky, F. M., Bishop, J. M. (2002). Short RNA duplexes produced by hydrolysis with Escherichia coli RNase III mediate effective RNA interference in mammalian cells. Proc. Natl. Acad. Sci. USA 99, 9942–9947.
Zimmermann, T. S., et al. (2006). RNAi-mediated gene silencing in non-human primates. Nature 441, 111–114.[CrossRef][Medline]
This article has been cited by other articles:
|
|
 |
|
  Y. J. Machida and A. Dutta The APC/C inhibitor, Emi1, is essential for prevention of rereplication Genes & Dev., January 15, 2007; 21(2): 184 - 194. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
