REVIEW

Yeast-based functional genomics and proteomics technologies: the first 15 years and beyond Bernhard Suter1, Daniel Auerbach2, and Igor Stagljar1

BioTechniques 40:625-644 (May 2006) doi 10.2144/000112151

Yeast-based functional genomics and proteomics technologies developed over the past decade have contributed greatly to our understanding of bacterial, yeast, fly, worm, and human gene functions. In this review, we highlight some of these yeast-based functional genomic and proteomic technologies that are advancing the utility of yeast as a model organism in molecular biology and speculate on their future uses. Such technologies include use of the yeast deletion strain collection, large-scale determination of protein localization in vivo, synthetic genetic array analysis, variations of the yeast two-hybrid system, protein microarrays, and tandem affinity purification (TAP)-tagging approaches. The integration of these advances with established technologies is invaluable in the drive toward a comprehensive understanding of protein structure and function in the cellular milieu.

INTRODUCTION Although the fields of functional genomics and proteomics are relatively young, they have benefited greatly from analysis of the baker’s yeast Saccharomyces cerevisiae (1). This single-celled fungus has long been an effective eukaryotic model system for understanding basic cellular processes due to its ease of manipulation and genetic tractability (2,3). Yeast has a short life cycle of 90 min, it is inexpensive to maintain and grow, it is stable in both the haploid and diploid state, and it is classified as a generally recognized as safe (GRAS) microorganism. Its haploid genome is of relatively low complexity (1.2 × 107 bp) and is packaged into 16 well-characterized chromosomes (4). Furthermore, yeast was the first eukaryotic genome for which genome sequence was reported (5), and its 6466 open reading frames (ORFs) (6,7) exist in a readily usable form (8). Annotated information on the function of these ORFs and their corresponding protein products is available through several databases, including the Saccharomyces Genome 1University

Vol. 40, No. 5 (2006)

Database (SGD; www.yeastgenome. org), the Yeast Protein Database (YPD; www.proteome.com), the Munich Information Center for Protein Sequences (MIPS) Comprehensive Yeast Genome Database (CYGD; mips. gsf.de/genre/proj/yeast/index.jsp), and the Yeast Resource Center (depts. washington.edu/~yeastrc). In addition, a big advantage of yeast compared with other eukaryotic model organisms is its relatively high rate of recombination between homologous DNA sequences that allows precise insertion of DNA sequences at specific locations within the yeast genome. This phenomenon enables a simple gene knockout technique to be performed in yeast using a PCR-mediated approach and the integration of a selectable marker and/or a mutant allele (9). Similar approaches can be used to introduce regulated promoters upstream of a given ORF (10) or to introduce different epitope tags (11). Yeast has also been successfully used for many years as a model for mammalian diseases and pathways. In fact, the cellular target of rapamycin, an immunosuppressant used as an anti-

rejection drug in tissue transplants, was first discovered in yeast and then subsequently verified in humans (12). Along these lines, yeast is a valuable predictor of human gene function; at least 31% of proteins encoded by yeast genes have human homologs, and conversely, nearly 50% of human genes implicated in heritable diseases have yeast homologs (13). In summary, yeast has definitively improved our understanding of fundamental cellular processes and metabolic pathways in humans and has facilitated the molecular analysis of many disease genes. In this review, we give a comprehensive overview of the most important yeast-based functional genomic and proteomic technologies that are advancing the utility of yeast as a model organism in modern biomolecular research. These approaches include utilization of the yeast deletion strain collection, large-scale determination of protein localization in vivo, synthetic genetic arrays, variations of the yeast two-hybrid system, protein microarrays, and tandem affinity purification (TAP)-tagging.

of Toronto, Toronto, ON, Canada and 2Dualsystems Biotech Inc., Zurich, Switzerland BioTechniques 625

REVIEW APPLICATIONS OF THE YEAST DELETION STRAIN COLLECTION In model organisms, the function of every known gene or predicted ORF can be analyzed by assessing growth defects or other phenotypes when the gene or ORF is deleted or the gene product is inactivated. Systematic generation of null or loss of function mutants and subsequent phenotypic analysis will therefore allow the functional categorization of all genes. Among eukaryotic model organisms, the yeast S. cerevisiae has been most extensively studied by systematic and genome-wide approaches.

Undoubtedly, one of the most important milestones for yeast functional genomics was the construction of a complete set of deletion mutants (14,15). This collection of null mutants allowed for the first time a comprehensive and parallel analysis of phenotypes in a model organism. In its final version, 5916 out of the originally annotated approximately 6200 yeast ORFs were successfully deleted for the construction of a heterozygous diploid collection that contains deletions in both essential and nonessential genes (14). In addition, sets of approximately 4800 nonessential deletions have been generated independently as haploid MATa and MATα strains or as

Pooling of deletion mutants - Drug

+ Drug Growth of mutants in presence or absence of drug

Amplification of barcodes

Hybridization to microarrays

Identification of barcoded deletions

Figure 1. Competitive growth assay for phenotypic analysis using the yeast deletion collection. Large pools of heterozygous or homozygous yeast deletion strains are grown in the presence Figure 1 or absence of a drug (pink squares). In the collection, each gene deletion (red and blue) is flanked by two sequences that contain unique barcodes (up-tag and down-tag). After DNA extraction, the barcodes are amplified using primers from conserved sequences in the tags and hybridized to barcode microarrays. Relative abundance of each bar-coded PCR product is compared between arrays from treated or untreated samples. Absence of a hybridization signal in the treated sample reveals sensitivity of the corresponding deletion strain (red deletion) to the drug. Note that this scheme depicts the hybridization of treated and untreated samples to different arrays. 626 BioTechniques

homozygous diploids. The collections are available from Research Genetics (ResGen) (www.resgen.com/products/ YEASTD.php3) or the Euroscarf consortium (web.uni-frankfurt.de/fb15/ mikro/euroscarf/col_index.html). Since the deletion collection was first described (15), numerous yeast laboratories have tested drug sensitivities and fitness defects of the mutant strains. Collection strains that are sensitive toward the cytotoxic or growth inhibitory effects of a specific compound can simply be identified by their absence or slow growth on medium containing the chemical. Plating assays have been used to test the homozygous diploid and haploid collections for sensitivities toward individual chemicals and different growth conditions (for some recent examples see References 16–20). In one extensive series of experiments, haploid gene deletion mutants were screened for hypersensitivities toward 12 inhibitory compounds (21). Drugs were selected that inhibited a set of different cellular pathways. These included, among others, (i) the microtubule-inhibitor benomyl; (ii) the immunosuppressants FK506 and cyclosporine that target calcineurinB; (iii) the ribonucleotide reductase inhibitor hydroxyurea; (iv) wortmannin, an inhibitor of phosphatidylkinase signaling; and (v) fluconazole, an antifungal compound that affects ergosterol biosynthesis. The screens identified a set of multidrug resistance genes that are required for resistance toward different compounds with diverse modes of action (i.e., PDR5, SNQ1, and YOR1). Filtering of these genes from the data sets revealed specific information about the pathways that are required to buffer the effects of the individual compounds. Increased compound sensitivities of individual mutant cells can be integrated into a chemical-genetic interaction profile. The different classes of mutant strains that are affected in presence of a compound generate distinct interaction profiles. Notably, chemical interaction profiling is also useful when a compound is not directed against a specific protein target. In this case, the interaction profile highlights primary pathways or cellular functions that are affected by the drug. Vol. 40, No. 5 (2006)

As an efficient method to monitor fitness defects, Ron Davis and coworkers established a protocol for competitive growth in large pools of deletion strains (14,15). This method allows massive parallel analysis of individual strains under different growth conditions in relatively small volumes of liquid culture (Figure 1). Growing a large pool of deletion strains in the presence of a specific bioactive compound, sensitive deletion strains are outcompeted by thousands of unaffected strains in the pool. A sophisticated system of unique DNA barcodes allows an unambiguous identification and quantification of individual mutants in the mixed population. Tags that contain barcode sequences are engineered into the 5′ and 3′ ends of each deletion cassette. Upon amplification of the tags, the presence or absence of each deletion strain is assessed by hybridization to barcode microarrays. Mutants that are sensitive to a specific drug are not detected by their corresponding barcodes, whereas barcode signals from unaffected or resistant strains are represented on the array. Strains that are slow growing under normal conditions and not specifically affected by the drug drop out in the untreated control sample and can therefore be recognized. Competitive growth experiments are better for the detection of weak sensitivities than direct plating assays. Furthermore, competitive growth requires only small amounts of liquid culture to screen the entire yeast collection and is therefore more cost-effective than direct plating assays when expensive compounds are tested. In the first release of the yeast deletion collection, the growth phenotypes of more than 500 deletion strains were assayed in parallel (15). Then, in the final assembly of the collection, Ron Davis and coworkers tested the entire homozygous deletion pool for changes under different growth conditions including pH, high salt, sorbitol, minimal media, galactose, and treatment with the antifungal compound nystatin (14). This work identified previously unknown genes that are required for growth under each tested condition, including 10 new genes that are required for growth in galactose. About Vol. 40, No. 5 (2006)

15% of the homozygous deletions were found to have a slow growth phenotype at 30°C in rich medium. Many of these genes encode ribosomal proteins and proteins involved in mitochondrial function and respiration. Mitochondrial genes were the focus of another experiment, in which the heterozygous and homozygous deletion collections were screened for growth on nonfermentable carbon sources (22). In total, 466 gene deletions were found to impair mitochondrial respiration, 201 of which encoded proteins with known mitochondrial localization or function. Human orthologs for yeast genes that have deletion phenotypes on nonfermentable carbon sources were then identified as candidate genes for human mitochondrial disease. This example shows how the conservation of cellular pathways among eukaryotes supports the use of yeast genomics to better understand these pathways in human. Many compounds that are used in cancer therapy mediate their effect through the induction of DNA damage. Notably, most of the currently used cancer therapeutics were originally introduced without their exact mode of action and the affected cellular pathways being defined. This includes, for example, the DNA cross-linking agent cisplatin, which is an effective chemotherapeutic for most germline testicular tumors (23). To understand the genetic requirements for drug sensitivity and resistance, an important concept developed by Lee Hartwell and Stephen Friend was to screen chemicals for toxicity in yeast cells that have metabolic defects similar to those frequently found in human cancers (24). Indeed, the high-throughput screening of the yeast deletion collection was successful in determining interaction profiles for different compounds that introduce structural DNA damage, including known anticancer drugs. Compounds and environmental agents that were screened include, for example, methyl-methanesulfonate and hydroxyurea (17), bleomycin (16), cisplatin (18), ionizing radiation (25), and UV radiation (26). These screens mainly uncovered genes that are involved in DNA repair, checkpoints, DNA recombination, and DNA replication. In the last example (26),

mutations in LSM1 and YAF9, which have human orthologs associated with cancer, conferred UV sensitivity. Recently, sensitivity in approximately 4700 homozygous deletion strains toward the induction of DNA damage was measured in a parallel analysis with 12 different compounds (27). A particular emphasis was placed on chemicals that induce cross-links between complementary DNA strands (e.g., the structurally related drugs cisplatin, oxaliplatin, and carboplatin). Clustering of the data showed the importance of nucleotide excision repair, but also of homologous recombination and postreplication repair in the cellular response toward these agents. It also highlighted a distinct profile for carboplatin, reflecting a different mode for damage response or reactivity of this drug. Furthermore, the genes MPH1, SHU1, SHU2, CSM2, and PSY3 were defined as a separate epistasis group that constitutes a separate branch in DNA repair by homologous recombination. While most of the screenings discussed in this review simply identified sensitivities and resistances of different mutant strains to drugs by testing growth and viability, a number of studies used the yeast deletion collection to study more complex phenotypes. Recent examples include two studies by the labs of Brian Kennedy and Stan Fields that used the yeast deletion collection to find factors that affect aging in yeast (28,29). In the first study, replicative life span (RLS), the measure of the number of cell divisions that can be performed by an individual mother cell, was determined for 564 deletion strains (28). Ten gene deletions were found to increase lifespan, and six of these genes were involved in nutrient response, specifically the target of rapamycin (TOR) and the Sch9 pathways. In the second study, chronological life span (CLS), a measure of the time nondividing cells can remain viable in liquid media, was assessed for the entire collection of homozygous diploid deletions (29). Mutations deficient in nutrient response and TOR signaling were also the most prominent in prolonging CLS. Consequently, removal of preferred yeast nitrogen sources glutamine and BioTechniques 627

REVIEW AA

B

of fitness defects in essential genes, which is not possible if homozygous diploid - Drug - Drug or haploid strains were Viable Viable used. On the other hand, the haploinsufficiency approach is limited when the drug - Drug - Drug target is not a protein Viable Viable (e.g., DNA damage) or when the direct effects of the drug are masked by a redundant protein function. Furthermore, + Drug + Drug Viable Viable growth deficiencies may not be detected in heterozygous deletion mutants where the relevant protein remains + Drug + Drug in excess of minimum Lethal Viable levels (Figure 2B) (22). An initial study used a small set of Figure 2. Drug-induced haploinsufficiency in heterozygous diploid strains. (A) A copy number reduction of a drug target gene (red 233 yeast deletions to circles) from two to one limits the amount of a potential drug target Figure 2 establish the feasibility and confers hypersensitivity to the drug (pink squares), whereas the drug has no effect in the presence of two functional gene copies. (B) of the haploinsufficency approach and the experHaploinsufficiency cannot be observed when the potential drug target is in excess even when one gene copy is deleted. imental procedures (30). More recently, two independent studies tested arginine, but also direct inhibition different compounds that are relevant of TOR signaling and inhibition of for human health and agriculture glutamine synthetase, led to an increase (31,32). The first study by Lum et al. in CLS. tested 78 compounds on a set of 3503 heterozygous yeast deletions (31). Drug-Induced Haploinsufficiency For a majority of the 20 compounds Screens with characterized functions that were included in the study, the correct targets The competitive growth approach and other components of the known is especially useful for the parallel target pathways were identified. Six analysis of drug-induced haploinsufof the correctly identified targets are ficiency in the heterozygous deletion involved in the ergosterol biosynthesis collection (30–32). A cell is haploinpathway, which is sensitive to various sufficient when lowering the dosage compounds (e.g., lovostatin and of a gene encoding a potential drug antifungal azoles). Furthermore, the target from two copies to one copy study also revealed a number of potential confers hypersensitivity to the drug new drug targets. For example, the drug (Figure 2A). The readouts from molsidomine, a potent vasodilator used haploinsufficiency screens are distinct in the treatment of angina (33) was from chemical-genetic screens with found to be an inhibitor of the Erg7p complete homozygous null mutations. lanosterol synthase. Heterozygotes Whereas screens with homozygous for the deletion of ERG7 showed deletions highlight genes that are hypersensitivity to molsidomine, and important for survival in the presence the purified lanosterol synthase was of a compound, the haploinsufficency inhibited in vitro by SIN-1, the first approach relies on a dosage effect and metabolic derivative of molsidomine. is expected to reveal the primary target This result revealed therefore the for the small molecule. Furthermore, likely molecular mechanism for the haploinsufficiency allows the analysis Haploinsufficiency

628 BioTechniques

No Haploinsufficiency

cholesterol-lowering properties of this drug in rats and humans. However, for some established compounds, no targets were identified, or the results from the screens were unexpected. The antimetabolite 5-flourouracil (5FU) is one of the most widely used chemotherapeutics for the treatment of solid tumors. Although 5FU was thought to inhibit thymidylate synthase (34), the deletion mutant of thymidylate synthase was not identified in the haploinsufficiency profile. However, a large set of genes involved in ribosomal RNA processing was sensitive to 5FU treatment. Subsequent characterization of ribosomal RNA (rRNA) intermediates by Lum et al. suggested that 5FU disrupts rRNA processing by the exosome complex (31). The second study by Giaever et al. (32) examined the complete collection of 5916 heterozygous deletion strains for sensitivity toward a set of 10 diverse compounds. Known interactions were supported for the anticancer drug methotrexate (DFR1), antifungal azoles (ERG11), and statin compounds that affect the HMG-CoA reductase (HMG1). Giaever et al. also identified the haploinsufficiency of rRNA processing mutants in the presence of 5FU, but attributed this phenotype to misincorporation of 5FU into RNA (32). Most interesting in this study, alverine citrate, a muscle relaxant, fenpropimorph, an antifungal, and dyclonine, an anesthetic, had similar profiles in the haploinsufficiency screens. Indeed, these three therapeutically distinct compounds were then found to share a similar chemical structure. This suggests that comprehensive haploinsufficiency profiling will possibly reveal unexpected relationships between activity and structure for other therapeutic compounds. Phenotypic analysis of the yeast mutant collections revealed invaluable clues about chemical genetic interactions and an overview of metabolic pathways that is not easily achieved in higher eukaryotes. The fact that many human disease-associated genes have conserved yeast counterparts (e.g., in DNA damage response) makes chemical genomics in yeast a powerful tool for testing relevant bioactive compounds (13). Unfortunately, many Vol. 40, No. 5 (2006)

oncogenes and tumor-suppressor genes that constitute potential drug targets have no clear orthologs in yeast. Orthologs of most human cancer genes can be found, however, in the genomes of metazoan model organisms like Drosophila melanogaster and Caenorhabditis elegans. Forward genetic screens have been traditionally more cumbersome in these organisms than in yeast. However, depletion of messenger RNA (mRNA) levels by RNA-mediated interference (RNAi) has been shown to be an efficient method to generate hypomorphic lossof-function phenotypes. RNAi can therefore be used for systematic and efficient phenome mapping in these organisms (35–37). RNAi libraries can also be generated to silence gene expression in mammalian cells (38). Hence, the parallel analysis of cell cultures that have the expression of different genes compromised by RNAi

strain and the deletion collection. Robotic high-density format pinning and replica printing of the deletion strains ensures high reproducibility and Synthetic Genetic Arrays and efficiency of the procedure. Results Genome-Wide Synthetic Lethal from the SGA analysis are then verified Screens by tetrad analysis or by a fast and more efficient growth analysis after random The synthetic genetic array (SGA) sporulation (for a more detailed methodology (Figure 3A) is an description see Reference 41). important application of the yeast SGA is especially good at showing deletion collection that allowed, for the functional redundancies that lead to first time, the systematic construction synthetic lethal (SL) or synthetic sick of double mutants and the subsequent (SS) interactions when two genes are analysis of growth phenotypes (39,40). compromised that have no or a lesser A query strain with a specific mutation effect on viability by themselves. SL (deletion or conditional allele) is mated or SS is known to occur when two with the entire haploid collection of the mutations compromise redundant opposite mating type. A tight selection pathways that are by themselves nonesprocedure allows only haploids, but not sential (Figure 3B). Alternatively, SL diploids, to germinate after sporulation. or SS interactions can be observed After the final selection step, phenowhen two mutations impinge on the types (e.g., sickness or lethality) are same essential pathway or when an scored in double mutants containing essential protein complex is destabiboth selective markers from the query lized. Previously, screens for synthetic lethality were A B C based on plasmid loss in A B C colony-sectoring assays (42). These approaches Viable - Drug Mating were hampered by practical Viable limitations, and only a handful of synthetic interactions were discovered in a single screen. Therefore, Sporulation Viable - Drug Viable with traditional synthetic lethal screens, the number of synthetic interactions found for a gene was Viable + Drug Haploid Selection always far from saturation. Viable The advent of SGA made it possible to analyze functional redundancies among yeast genes in a Double Mutant Selection Lethal + Drug comprehensive manner, Lethal although some false negatives (real interactions that were not identified) are still expected to occur. An Figure 3. Screens for synthetic genetic and chemical genetic interactions. (A) Synthetic lethal screening using syninitial SGA study revealed Figure deletion 3 thetic genetic arrays (SGA). A haploid strain containing the query mutation (red) is combined with the arrayed collection of the opposite mating type (blue). After mating, diploids are selected and sporulated to yield the haploid 291 interactions among progeny. For the crucial selection step against diploids, the query strain contains a special construct that allows for the 204 genes from only 8 selection of haploids from one mating type. The use of additional resistance markers (e.g., canavanine resistance, can1Δ) different queries (39). This makes the process more stringent, specifically allowing additional selection against diploids that became homozygous by number was then expanded gene conversion. The final selection for haploid double mutants is done with the antibiotic resistances from the deletion collection and the query strain. In case of a synthetic sick or lethal interaction, the double mutant is compromised or can- by crossing 132 selected not be recovered. (B) The principle of synthetic lethality. Inactivation (deletion) of two genes (red and blue) in redundant query mutations that pathways leads to loss of viability, whereas inactivation of either one gene has no effect. (C) Drug-induced inhibition of revealed a synthetic intergrowth in haploid or homozygous diploid deletion strains (chemical-genetic interaction). A protein product of a gene (red action network including circles) is inactivated by treatment with a drug (pink squares). The protein product of a second gene (blue circles) is by approximately 1000 itself not essential but prevents loss of viability in the presence of the drug. Deletion of the second gene leads therefore to hypersensitivity to a dose of the drug that that is not lethal in a wild-type cell. genes and approximately Vol. 40, No. 5 (2006)

could allow rapid assessment of drug effects in more complex cells.

BioTechniques 629

REVIEW 4000 interactions (40). The inference from clustered SGA data allows the assignment of functions to previously unknown genes (40). For example, the clustering of synthetic interactions and follow-up studies revealed that Csm3p cooperates with Mrc1p and Tof1p in the DNA replication checkpoint. Furthermore, focused SGA screens have been applied to address specific questions in several research areas, including polarized growth (43), sister chromatid cohesion (44,45), DNA replication (46–48), chitin synthesis for the cell wall (49), and transcription (50). The SGA screens that were done so far detected only a fraction of the synthetic interactions that are presumed to occur in a yeast cell. However, the massive increase in the number of genetic interaction data will eventually allow a comprehensive overview of the genetic interaction network. Approximately 20% (or 1105) of yeast genes are essential and therefore not represented in the haploid deletion collection and not accessible for SGA. Recently, methods have been developed that make essential genes accessible for high-throughput analysis. In one approach, a collection of promoterreplacement alleles was constructed for over two-thirds of all essential yeast genes (10). In these strains, an activator binding sequence is integrated into the promoters of the candidate genes conferring high expression levels to the protein in the presence of the activator (tetA). Inhibition of tetA by the chemical doxycycline leads to a shutoff of target gene expression coming close to the effect that would be expected for a deletion of the essential gene. Upon addition of the inhibitor and depletion of mRNA levels, the strains in the collection were analyzed for cellular morphology, size, drug sensitivity, and mRNA expression profiles. Importantly, the promoterreplacement collection is also suitable for SGA analysis. More recently, 30 different temperature-sensitive mutants or deletions in nonessential genes with defects in diverse cellular processes were crossed with an array containing 575 promoter-replacement alleles in essential genes (51). The resulting synthetic interaction network showed a higher number of synthetic 630 BioTechniques

lethal interactions for essential genes, compared with nonessential genes. The promoter replacement system was also used to assign functions to previously uncharacterized genes. This included the identification of Pga1p as a new factor that has a role in protein sorting or modification in the endoplasmic reticulum (51). Finally, an effort is currently underway to generate a comprehensive collection of temperature-sensitive mutations in essential genes as a resource for future SGA analysis (C. Boone, personal communication). A modified version of SGA has been developed that allows quantitation of aggravating or alleviating interactions between pairs of genes with categorized functions by using epistatic miniarray profiles (eMAPs) (52). This eMAP strategy has been successfully used for cellular trafficking and secretory processes. The same study also introduced destabilization of mRNA by disruption of the 3′ untranslated region (UTR) sequences as a method to study phenotypes that arise by reduction in the level of essential gene expression. An alternative methodology to assess synthetic interaction between genes is synthetic lethality analysis by microarray (SLAM), which combines the principles of SGA and competitive growth. The query genes are integrated into the haploid deletion (SLAM) by transformation (53). To assess the quantitative strength of each interaction, barcodes are then used to detect the relative abundance of individual deletions on microarrays. SLAM was successfully used to analyze functional relationships between helicases that are involved in chromosome metabolism. Recently, a modified version of SLAM has been developed (dSLAM), in which heterozygous diploid pools are transformed with the query and haploids are selected before microarray analysis (54). The application of SGA has also been extended from the simple combination of mutant alleles (deletions or conditional mutants) to plasmid-based assays, looking at how overexpression of genes affects viability in certain mutant backgrounds. Overexpression of a given gene can either rescue viability in a mutant background (multicopy suppression) or additionally

compromise cell growth (dosage lethality). Dosage lethality screening takes advantage of a collection of 5800 cloned yeast genes, driven by the GAL1 promoter, which was originally generated for the production of protein arrays (55). Recently, a comprehensive approach used SL and synthetic dosage lethality (SDL) to identify genes that are involved in chromosome segregation (44). Whereas both the SL and SDL screens identified genes with functions in chromosome maintenance, they were nonoverlapping for many individual genes. This can be explained by distinct effects on cellular pathways associated with loss of function (deletion) and gain of function (gene overexpression). Besides known chromosome segregation factors, the SDL and SL screens also uncovered the iron transcription factor Rcs1p as a new component of the chromosomal segregation machinery. Very recently, gene dosage phenotypes were systematically analyzed in 5280 yeast strains overexpressing individual genes (56). Approximately 15% of yeast genes conferred toxicity in the wildtype background when overexpressed. Furthermore, gene overexpression in the absence of the cyclin-dependent kinase Pho85p revealed targets for phosphorylation by Pho85p, among them the yeast calcineurin-responsive transcription factor Crz1p. The principles of SGA analysis can also be extended and combined with other approaches. Drug sensitivities can be viewed as a form of synthetic genetic interactions (chemical genetic interactions) (Figure 3C). Charlie Boone and coworkers demonstrated how results from drug sensitivity and SL screens can be combined in chemical-genetic interaction profiles (21). Ideally, in such a chemicalgenetic profile, deletion strains that are sensitive to a particular compound are also synthetically lethal or sick with a mutation in the compound’s target gene. However, since the effects of most drugs are pleiotropic and the inactivation of a target by a drug is not completely equivalent to a gene deletion, the genetic interaction profile of a specific drug is typically not completely overlapping with the interaction profile of the drug target gene. Indeed, the same study demonstrated Vol. 40, No. 5 (2006)

similar but not identical interactions for benomyl and its target TUB2, for fluconazole and its target ERG11, and also for FK506, cyclosporine, and their target calcineurinB (21). In the future, the SGA methodology will also be used to obtain information in addition to growth phenotypes and to integrate those into the yeast interaction network. So far, viable double mutants, generated in SGA screens, have been analyzed for growth and then discarded. However, it would be desirable to keep these strains as a valuable resource for further followup studies. A more recent effort aims now at the characterization of cellular morphologies of double and single mutants (C. Boone, personal communication). Looking beyond yeast, some of the principles for synthetic lethal screens by SGA can also be adapted for higher eukaryotes. Systematic screens using gene silencing by RNAi have been performed in the worm C. elegans (36,57,58). This method can be easily adapted for synthetic lethal screens by simultaneously administering two different RNAi species. GENOME-WIDE ANALYSIS OF PROTEIN LOCALIZATION One aspect that is critical for the understanding of protein function is protein localization within distinct subcellular compartments and microenvironments. A first large-scale localization study in S. cerevisiae determined the subcellular localization of 2744 yeast proteins (59). This study relied on transposon-mediated random tagging with hemaglutinin (HA) and paramyxovirus SV5 epitopes and overexpression of the tagged proteins. Although this work was very successful in showing the subcellular localization of a large amount of protein, errors may be introduced by the strategy of random tag insertion or by protein overexpression interfering with localization signals. Furthermore, the immunostaining procedure required fixation and processing of the cells. Therefore, a very important improvement toward a systematic localization of proteins constituted the use of light microscopy along with genetically encoded fluorescent proteins that Vol. 40, No. 5 (2006)

allowed the visualization of protein trafficking and localization in vivo. In a large-scale localization study using green fluorescent protein (GFP) that was preformed by the O’Shea and Weissman laboratories, a GFP tag was inserted at the C terminus of 6029 yeast ORFs (60). The results of these studies are accessible at yeastgfp.ucsf. edu. Out of this collection, 4156 strains expressed the GFP-tagged protein at detectable levels, and therefore 75% of the yeast proteome was covered. The GFP-tagged proteins were assigned to 23 localization categories that are refined for subcellular compartments and organelles (e.g., cytoskeleton, nucleolus, nuclear membrane, etc.). The comparison of data published by Huh et al. (60) with those from earlier localization studies revealed an agreement of 80%, supporting the reliability of the data. A potential source of error is introduced by the C-terminal fusion of the GFP protein, which may cause mislocalization through steric hindrance or interruption of C-terminal localization/retention signals. An additional, more recent, study by Kohlwein and colleagues investigated the localization pattern of proteins involved in lipid metabolism in S. cerevisiae by GFP tagging and high-resolution confocal laser-scanning microscopy (61). Enzyme localization patterns for phospholipids, sterol, and sphingolipid biosynthetic pathways indicated concerted regulation of sterol and sphingolipid metabolism in the endoplasmic reticulum, lipid droplets, vesicles, and the Golgi apparatus. An important method to complement and to verify GFP localization data are enrichment or isolation of cellular organelles and subsequent identification of proteins by mass spectrometry. For example, from the 16 components of the spindle pole body that were identified by mass spectrometry (MS) (62), 14 were visible by GFP and localized to the spindle pole (60). Similarly, GFP localization to the nuclear membrane was also confirmed for most of the proteins that were found in the purified nuclear pore complex (63). The GFP localization information can be combined with data from other large-scale studies to confirm and extend predictions about protein functions. For example, proteins

with a common localization that are also grouped together by genetic or physical interactions are very likely to have a common function. These predictions are especially useful in the case of proteins for which little functional data exist. In the future, a possible extension of the GFP localization studies could be the introduction of GFP fusion proteins into the deletion collection using highthroughput strain construction by SGA. Effects of individual gene deletions on protein localization can then be analyzed, allowing a genetic dissection of complex protein localization patterns within the cell. YEAST INTERACTOME ANALYSIS BY TAP-TAGGING AND MASS SPECTROMETRY A major focus for functional proteomics today is the analysis of physical interactions between proteins and the organization of the cellular proteome into multiprotein complexes. Biochemical isolation of these complexes, followed by identification of their individual components, is highly predictive for functional associations among proteins. By the analysis of molecular environments, new insights can also be obtained for proteins with an already known function. Thus, one of the major advancements in functional proteomics today is the systematic isolation of protein complexes followed by the identification of their components by mass spectrometry. Ideally, a method for large-scale protein purifications must be highly discriminating against nonspecific protein background but retain the specific components of the complex. Furthermore, the method should be generic to deliver reproducible and comparable results. The feasibility of high-throughput protein interaction studies depends therefore on protein tags that can be recognized by specific antibodies or by other high-affinity interactors. In 2002, two interactive proteomics papers were published that generated extensive protein-protein interaction networks (64,65). The Tyers group used a onestep purification procedure using the FLAG® epitope tag as their method of BioTechniques 631

REVIEW Integration of TAP cassette at ORFX ORFX::TAP

Growth of large cultures Protein extraction

Binding to IgG beads

IgG beads

Protein A TEV CBP Bait

TEV IgG beads

Protein A

Cleavage by TEV protease CBP Bait

Calmodulin beads

Ca2+ Ca2+

Binding to calmodulin beads CBP Bait

Ca2+ Ca2+

Calmodulin beads

Ca2+

Release from calmodulin beads CBP Bait

Ca2+

Separation of proteins by SDS-PAGE

Identification of peptides by MALDI-TOF and database searches

Excision of bands

Trypsin digest

m/z

Figure 4. Overview of the tandem affinity procedure (TAP). The TAP-tag of a protein A Figure consists 4 domain and a calmodulin binding protein (CBP) that are separated by a recognition sequence for the tobacco etch virus (TEV) protease. The TAP sequence is integrated in-frame at the C terminus (or N terminus) of a gene. Protein extracts are made from large cultures of cells expressing the TAP-tagged proteins. In a first purification step, the fusion proteins bind to immunoglobulin G (IgG) beads by their protein A domain. Cleavage by the TEV protease then releases the fusion proteins and the associated proteins or protein complexes from the IgG beads. In a second purification step, the fusion proteins bind to calmodulin-coated beads by their CBP domain in the presence of Ca2+. After additional washing, the tagged proteins are eluted by the addition of the chelator EGTA that binds to Ca2+. Purified protein complexes are then separated by denaturing gel electrophoresis. Distinct protein bands are excised and digested into small peptides with trypsin. The peptides are then identified by mass spectrometry, and the identity of the proteins is determined by database searches. m/z, mass-to-charge ratio; SDS-PAGE, sodium dodecyl sulfate polyacrylamide gel electrophoresis.

choice (64). Bait proteins (725) were screened, among them a large number of protein kinases, phosphatases, and proteins implicated in DNA damage response. In total, 3617 protein interactions were observed with 493 baits. The overexpression of the FLAG-tagged protein is a potential advantage for the detection of a large number of interaction partners, but on the other hand, may also result in a higher amount of nonspecific interactions. 632 BioTechniques

The TAP method allows purification of protein complexes under largely physiological conditions (66–68) (Figure 4). This methodology is highly discriminating against nonspecific protein background, because it relies on the utilization of two tags instead of one. Furthermore, the TAP-tagged protein is expressed from its normal chromosomal locus, and therefore, possible artifacts by protein overexpression are avoided. The tagged protein is isolated along with associated proteins or

protein complexes by sequential purification over two high-affinity columns (Figure 4). In the first column, the proteinA moiety of the TAP-tag binds to immobilized immunoglobulins. After release from the column by protease cleavage using the tobacco etch virus (TEV) protease, the complex is bound to calmodulin-coated beads via the calmodulin binding peptide (CBP) tag. The removal of calcium ions with EGTA elutes the purified protein complexes. The proteins are then separated by gel electrophoresis and identified by matrixassisted laser desorption/ionization time-of-flight MS (MALDI-TOF MS). This TAP/MS-based approach is also readily applicable for the retrieval of protein complexes from human cell lines (65,69). The described purification scheme was successfully used by the biotech company Cellzome (Heidelberg, Germany) in an initial large-scale proteomic study in which 589 tagged proteins were purified from S. cerevisiae (65). Seventy-eight percent of the purified proteins were found with associated partners forming 232 multiprotein complexes. Very recently, Gavin et al. published a follow-up on the TAP approach that covered the entire yeast proteome with all ORFs tagged. In total, 1993 TAP-fusion proteins were purified, and 88% of these were identified with at least one partner (68). One major concern is how the TAP-tag may interfere with complex assembly and protein function or localization. Indeed, for the first study, in 18% of the cases when essential genes were tagged at their C terminus, no viable strains were obtained (65). An example for interference with complex assembly was observed for the CCT (chaperonincontaining tailless complex polypeptide 1) complex where the C termini of the eight subunits lie on the interaction surfaces in the ring-like core (68). This loss of function may be circumvented by tagging at the N terminus. Despite these pitfalls, isolation of multiprotein complexes by TAP-tag has become a standard method to show biochemical interactions in yeast. In an independent study, the Weissman and O’Shea groups fused all ORFs in the yeast genome with C-terminal TAP-tags and assessed the expression level of each individual Vol. 40, No. 5 (2006)

tagged protein (70). Abundance ranged from fewer than 50 to more than 106 molecules per cell, and 80% of the yeast proteome was expressed under normal growth conditions. This collection was used to direct another large-scale approach that involves the systematic purification of all yeast proteins and the subsequent identification of complex components by MALDI-TOF or liquid chromatography followed by tandem MS (LC-MS/MS) (71). Although a large number of protein complexes have been identified using the TAP-tag, there is still a considerable potential to improve the analysis of the yeast proteome. Being very discriminative in excluding nonspecific protein binding, the TAP-purification method may fail to detect transient interactions or low stoichiometry complex members. Furthermore, for TAP purifications followed by MS, a considerable volume of cells is used, and the preparation process is very labor-intensive. This is a disadvantage, especially when large numbers of protein-protein interactions have to be studied under a variety of environmental conditions. Hence, it would be a great advancement if protein-protein interactions could be assessed in a massive parallel approach, comparable to large-scale genetic and

two-hybrid methods. Ideally, such a method would rely on a directed coimmunoprecipitation (co-IP) of differently tagged proteins, which allows a small sample size for cell growth and processing. Finally, co-IP could also be performed under less stringent conditions to favor the detection of transient and weak interactions. Nevertheless, some proteins will remain relatively inaccessible for biochemical isolation or refractory to tagging. These include membrane protein complexes that are difficult to analyze biochemically but often constitute important drug targets. Recently, the TAP-tag protocol was adapted for the identification of membrane protein complexes. This allowed retrieval of 340 of 628 TAP-tagged membrane proteins (68). Although several membrane-protein complexes were identified this way, including the Q/t-SNARE complex, alternative methods are still required to get a better insight into membraneassociated complexes. To identify protein interaction partners of membrane proteins, a two-hybrid-based screening method has been developed to monitor these interactions indirectly (see the section on membrane two-hybrid screens). Undoubtedly, additional approaches are required to improve our

Figure 5. Identification of protein interactions on proteome chips. Proteins from an expression library, but also peptides or antibodies, can be printed on the surface of glass slides. The array can be probed for binding of diverse molecules, including proteins, DNA, lipids in liposomes, and small molecules. Interactions can be detected by a fluorescent label. Alternatively, radioactive probes, mass spectrometry, and atomic force microscopy (AFM) can be used. Vol. 40, No. 5 (2006)

understanding of protein-protein interactions and to cover the entire complexity of the yeast proteome. PROTEIN MICROARRAYS AND APPLICATION ON YEAST PROTEINS The protein microarray technology is an important development for high-throughput analysis of entire proteomes and is expected to transform biochemical analysis procedures in the future. Protein microarrays are generally of two types: analytical (or affinity) arrays and functional protein arrays. Analytical arrays contain antibodies, or other relevant protein binding entities, and are very useful for monitoring protein expression levels and for clinical diagnostics (72). For functional protein microarrays, a set of proteins is deposited onto the surface of a slide under nondenaturing conditions and characterized using a wide range of biochemical assays. Binding assays for the immobilized proteins can be carried out using either a biotinylated probe, which can be detected by fluorescencelabeled streptavidin, or a probe that has been labeled directly with fluorescence. Functional protein arrays that cover a whole proteome can detect binding of a variety of labeled substrates, including protein ligands, lipids, small molecules, and DNA (see Figure 5) (reviewed in Reference 73). MacBeath and Schreiber first demonstrated the feasibility of protein microarrays in screening for proteinprotein interactions and in identifying the substrates of protein kinases and the protein targets of small molecules (74). Important work by the Snyder laboratory highlighted the potential of protein arrays for large-scale biochemical analysis in yeast. The overexpression of 5800 ORFs tagged with glutathione S-transferase (GST)fusion protein allowed the construction of the first full-scale proteome microarray (55). The GST-fusions were purified in a high-throughput protocol, and detectable proteins were printed onto slides to form a yeast proteome microarray. The experiment then identified 6 known and 33 putative new binding partners for calmodulin BioTechniques 633

REVIEW A

events. The study found that distinct sets of substrates were phosphorylated by each kinase and that most substrates were only phosphorylated by a small X number of kinases, indicating a strong DBD preference of individual kinases for their respective targets. The authors HIS3 confirmed that the phosphorylation lacZ status of a number of proteins, for which kinases were identified by this in vitro approach, indeed depended on B B the presence of the respective kinases in vivo. In another paper, the Snyder Y group demonstrated that compreAD hensive proteome arrays are suitable to examine antibody specificities (76). Protein arrays were used to probe HIS3 the yeast proteome with 14 different lacZ commercial antibodies that identified the correct cognate proteins but also showed antibody cross-reactivity with C other yeast proteins. Proteome arrays C have also been probed for DNA binding activities with fluorescent-labeled yeast genomic DNA (77). In total, 200 D A Y X proteins were identified, half of which DBD were not known to bind DNA previously. Surprisingly, a DNA binding HIS3 lacZ activity was discovered for a metabolic enzyme, Arg5,6p, which is involved in Figure 6. The yeast two-hybrid system. (A) To ornithine biosynthesis (a precursor to construct a bait in the yeast two-hybrid system, a It seems very likely that such 6 protein of interest X is fused to the DNA binding Figurearginine). unbiased proteomic approaches using domain (DBD) of a transcription factor. When expressed on its own in yeast, the bait will not proteome arrays will also uncover activate transcription since it lacks a transcripunexpected new functions for other tional activation domain (AD). (B) Likewise, a eukaryotic proteins in the future. prey is constructed by fusing a second protein Perhaps the most important appliof interest Y to the AD of a transcription factor. cation of the proteome microarray The AD-Y fusion is unable to activate transcription on its own, since it is not situated near a technology is the screening for small promoter. (C) Co-expression of the interacting molecule targets. The binding of DBD-X and AD-Y fusion proteins reconstitutes bioactive compounds to individual a functional transcription factor situated at a proproteins on the array can identify drug moter. Consequently, the reporter gene located downstream of the reporter is activated, and the specificities, but also highlight potential protein-protein interaction between the proteins side effects. One important study X and Y is measured using the product of the exemplifies how proteome chips can be reporter gene. Common reporter genes in yeast used to identify compounds that modify two-hybrid systems include auxotrophic growth the cellular signaling in response to markers, such as the HIS3 or ADE2 genes, or a color marker, such as lacZ. the anticancer drug rapamycin (78). Rapamycin specifically targets the TOR kinases that regulate cell growth and a number of proteins that interin response to nutrient signals (12,79). acted with the lipid phosphoinositide. First, a chemical suppressor screen Recently, the yeast proteome was used to screen for compounds microarray technology was used to that enhance or inhibit the effect of determine the substrate specificities the anticancer drug rapamycin on of 87 different yeast kinases (75). The yeast cell growth. Biotinylated forms tested kinases phosphorylated 1325 of the molecules that were found to different protein targets in a total of completely suppress the effect of approximately 4200 phosphorylation rapamycin in the screen were then

A

634 BioTechniques

used to probe yeast proteome arrays for direct interactions. Putative target proteins identified by this way included Tep1p, a homolog of the mammalian tumor suppressor PTEN, and Nir1p, which is a new component of the TOR signaling network. This study provides an example of how the combination of chemical genetics with proteome arrays could uncover potential new drug targets. Another possible strategy to assess drug target specificity could consist, for example, of a combination of yeast proteome array and haploinsufficiency screens. Hence, yeast proteome arrays can be used to discover potential new drug targets but also to validate targets that have been discovered by other approaches. Right now, development of commercially available proteome arrays is still in its infancy, and important technical issues remain unresolved. However, given the versatility of this approach in addressing molecule binding specificities and other functional aspects of entire proteomes, a major role for this technology in drug development can be expected in the near future. YEAST TWO-HYBRID SYSTEM Originally devised by Fields and Song, the yeast two-hybrid system uses transcriptional machinery of S. cerevisiae as a screening tool to discover novel protein interactions (80). In the original implementation of the yeast two-hybrid system, a known protein of interest, the bait, is expressed as a fusion to a DNA binding domain of GAL4 (Figure 6A), and an interacting protein, the prey, is expressed as a fusion to the GAL4 transactivation domain (Figure 6B). When expressed on its own, neither the bait nor the prey activates the reporter genes integrated into the yeast genome. However, when bait and prey are co-expressed, their interaction reconstitutes a functional transcription factor that is situated upstream of the reporter genes. This two-hybrid transcription factor activates the reporter genes, whose output is measured, for instance, as growth of yeast clones on selective medium or as blue coloration in a β-galactosidase assay (Figure 6C). Vol. 40, No. 5 (2006)

Traditionally, the yeast two-hybrid system has been used by researchers to tackle a single protein of interest (see Reference 81 for an in-depth description of a typical yeast two-hybrid screen). Novel potential interactors resulting from the screen are then confirmed by independent methods, such as in vitro pull-down assays or co-immunoprecipitation. Confirmation of interactions found in the yeast two-hybrid system is needed, since overexpression of foreign proteins in yeast often yields spurious, nonspecific interactions (so-called false positives) (see Reference 82 for a review on false positives in the yeast two-hybrid system). Once the candidate interactions have been confirmed, most researchers then move on to investigate the functional significance of these interactions in a biological model system. Eventually, they may come back to carry out another yeast twohybrid screen, for example by using one of the interactors found in the original screen as a bait. Albeit being a slow affair, this low-throughput screening has yielded invaluable information on protein interactions over the past decade. In fact, over 50% of all interactions published in Medline are derived from yeast two-hybrid screens (83). Attention has recently shifted toward trying to determine protein interactions on a large scale. These studies, termed high-throughput screens or interaction maps do not focus on a single protein of interest, but instead try to use available genome information to create a set of baits encompassing the entire expressed genome of an organism. As this approach deals with hundreds or even thousands of baits and preys, some form of automation is needed. Luckily, the yeast two-hybrid system is very amenable to robotic platforms, and several schemes have been developed to automate a yeast two-hybrid screen from cloning down to analysis of the interactors (84). Initial attempts at creating genome-wide protein interaction maps focused on organisms with relatively few genes, such as viruses (85,86) and prokaryotic species (87). The first real attempt at mapping the complete interactome of a eukaryote was carried out for S. cerevisiae. Two independent groups created sets of baits and preys from the majority of the Vol. 40, No. 5 (2006)

6000 predicted genes of S. cerevisiae and used these in exhaustive large-scale screens (84,88). What became rapidly apparent when analyzing the two high-throughput studies carried out for yeast was that the generated interactomes were far from complete. Analysis of the yeast interaction networks suggests that on average, every protein interacts with at least 2–3 partners. Assuming that the 6466 ORFs in yeast give rise to an equal number of proteins (which is a conservative estimate, since it does not take into account additional diversity through posttranslational modifications), one would expect to have at least 12,000–18,000 possible protein interactions in yeast. The considerably lower number of interactions found in the two systematic large-scale screenings carried out to date suggests that the false negative rate (i.e., a protein interaction that is undetectable in a yeast two-hybrid assay) is rather high. This is also obvious when looking at the overlap between large-scale screens carried out using the same organism. Comparison of yeast interaction data derived from the Uetz and Ito publications shows that only 20% of all protein interactions are found in both data sets, suggesting that both screens were far from saturating or contained false positives (84,88). The recent large-scale screens carried out with the metazoan organisms D. melanogaster, C. elegans, Homo sapiens, and Plasmodium falciparum (89–94) are likely to face even greater challenges when trying to cover the entire interactome, since extensive alternative splicing and posttranslational modifications massively increase

AA Y AD

DBD

HIS3

BB Y AD

DBD

HIS3

C

C Y

AD

DBD

HIS3

Figure 7. Small molecule yeast two-hybrid screening. (A) A scaffold bait is constructed by fusing a DNA binding domain (DBD) to a small Figure 7 molecule binding protein such as dihyrofolate reductase (DHFR). Simultaneously, each yeast cell expresses a particular activation domain (AD)-fused prey from a cDNA library. (B) A hybrid compound consisting of a small molecule covalently linked to methotrexate is added, which crosses the yeast cell membrane and binds to the DBD-DHFR bait via its methotrexate part. In this way, the other part of the small molecule is displayed by the scaffold bait. (C) If the ADprey binds to the small molecule displayed from the scaffold bait, a functional transcription factor is reconstituted via the small molecule-protein interaction, resulting in activation of the downstream reporter gene.

Table 1. Large-Scale Yeast Two-Hybrid Screens Performed Thus Far Organism Helicobacter pylori Saccharomyces cerevisiae Drosophila melanogaster

Genes

Interactions Reported

Reference

1590

1280

87

~6000

967

84

4549

88

20,405

89

1814

93

~14,000

Caenorhabditis elegans

~20,000

4027

90

Plasmodium falciparum

5300

2864

94

~30,000

2800

91

3186

92

Homo sapiens

BioTechniques 635

REVIEW the complexity of the proteome. Very recently, two independent large-scale yeast two-hybrid screens have been published that provided for the first time an overview of the human interactome (91,92). Thus, methods are urgently needed that allow comprehensive screening of large arrays of baits and preys against each other, such as the recently developed two-phase mating technique (95). Table 1 presents a selection of some large-scale yeast two-hybrid studies carried out to date. Yeast Two-Hybrid System in Drug Discovery As can be expected from a system that is so powerful for characterizing protein interactions in a natural setting, the yeast two-hybrid system has also been applied to the arduous task of drug discovery. In one adaptation, which has been termed small molecule three-hybrid, the two-hybrid system is modified to detect small moleculeprotein interactions instead of proteinprotein interactions. To achieve this, a bait is constructed by fusing the DNA binding domain to a protein with an affinity for a defined small molecule (Figure 7A). For example, dihydrofolate reductase (DHFR) binds the small molecule methotrexate. In a second step, the compound under investigation is chemically coupled to methotrexate,

creating a hybrid molecule. When added to yeast expressing the DNA binding domain-DHFR bait, the methotrexate hybrid compound binds to the bait (Figure 7B). In the final step, the yeast is transformed with a cDNA library encoding activation domain fusions, and those proteins that bind to the displayed hybrid molecule are selected (Figure 7C). Several different versions of this small molecule three-hybrid approach exist and have been shown to work in proof-of-concept studies (96), as well as for screening for binding partners of a given compound (97). Perhaps the most interesting application to date has been its use in assaying the specificity of cyclin-dependent kinase inhibitors (96). Overall, this type of approach shows great promise in characterizing the specificity of therapeutically relevant compounds in an in vivo like setting. As a genetic selection assay for protein-protein interactions, the yeast two-hybrid system can also be manipulated to screen for inhibitors of a protein-protein interaction. Here, the selection mechanism is reversed, such that the interaction between a bait and its cognate prey leads to activation of a toxic reporter gene, which then kills the yeast when grown on a special selective medium (Figure 8A). The addition of a compound, which inhibits the protein interaction, prevents expression of the toxic gene and restores yeast growth on

A

Y

AD

A X

Cell death

DBD HIS3

B

B Y

AD X DBD

HIS3

636 BioTechniques

Rescue from cell death

Figure 8. Screening for compounds that inhibit a protein-protein interaction. (A) An interacting protein pair is expressed as the fusion of DBD-X [a protein of interest X fused to the DNA binding domain (DBD)] and AD-Y [a protein of interest Y fused to the activation domain (AD)]. The reconstitution of an artificial transcription factor in the yeast two-hybrid system activates the downstream reporter gene, which converts a compound added to the medium into a toxic end product, resulting in yeast cell death. (B) Addition of a compound, which interferes with the protein-protein interaction, prevents reconstitution of the hybrid transcription factor. Consequently, the reporter gene is not activated, and the yeast grows on selective medium.

the selective medium (Figure 8B). The most commonly used reporter gene is URA3, which converts the compound 5-fluoroorotic acid (5FOA) into a toxic metabolite, resulting in growth arrest and cell death (98). Thus, a collection of small molecules can be screened to identify those candidates that inhibit a given protein interaction. This reverse yeast two-hybrid system has been adapted to a high-throughput format and has been used to demonstrate the disruption of activin receptor R1 binding to the immunophilin FKBP12 by the small molecule FK506 (99). An alternative approach to screening using the yeast two-hybrid system was used by Khazak and coworkers; here, disruption of a protein-protein interaction by small molecules is detected by measuring the output of two colorimetric reporter genes. Using the interaction between Ras and Raf-1, the authors identified a compound which, when tested in cell culture, was able to cause reversion of ras-transformed phenotypes including morphology, in vitro invasiveness, and anchorageindependent growth (100). Tackling More Complex Problems: Modification-Dependent and Compartment-Dependent Protein Interactions In the past decade, it has been demonstrated that the yeast two-hybrid system can tackle a surprisingly broad range of different protein interactions (101). However, it has also become clear that certain classes of proteins and protein interactions are outside of the scope of yeast two-hybrid. Unfortunately, precisely those protein interactions become increasingly important when one moves toward establishing protein interaction maps of higher eukaryotes such as mouse or human or when trying to take advantage of protein interactions as novel avenues in drug discovery. For example, yeast has a much more limited repertoire of posttranslational modifications than higher eukaryotes, such as C. elegans or humans. Therefore, many protein interactions that are dependent on such modifications will be undetectable in a yeast two-hybrid screen since yeast lacks the necessary Vol. 40, No. 5 (2006)

modification-dependent interactome map. Another limitation of yeast two-hybrid touches ������� integral membrane proteins. In the yeast two-hybrid �� ������ �� system, both the bait and ��� � the prey have to be localized in the nucleus to ensure detection of the interaction. Integral membrane proteins cannot be used as baits or preys since they would be � � B immediately translocated to the membrane during translation and therefore � ���� ������� would not be able to reach the nucleus. Thus, most ������ �� interactions involving ���� integral membrane proteins will be undetectable by yeast two-hybrid. In recent years, novel screening Figure 9. The reverse Ras system. (A) An integral membrane assays, based on the yeast protein or a membrane-associated protein is expressed as a bait in two-hybrid system, were a yeast strain carrying a mutation in the cdc25 gene. The cdc25-2 generated that aim at mutation prevents growth of the yeast strain at the nonpermissive the detection of protein temperature of 36°C. (B) Co-expression of an interacting protein fused to a constitutively active ras mutant (mRas) targets the interactions outside of the mRas-Y fusion to the membrane, where it acts to complement the nucleus. These approaches cdc25-2 mutation, leading to yeast growth at 36°C. include among others the son of sevenless (SOS) recruitment system (103), machinery to modify the exogenously an interaction assay based on RNA expressed bait or prey proteins. A polymerase III activation (104), the possible solution to this problem is to use of G protein fusions (105), and a co-express the modifying enzyme along system based on the unfolded protein with its substrate. Fields and collabo������ � response (106). Here, we will focus on rators have recently presented a tethered the reverse Ras (rRas) system (107) and catalysis two-hybrid system, which the split-ubiquitin membrane-based they successfully used to map interacyeast two-hybrid system (MbYTH tions dependent on phosphorylation and system), since they have seen the acetylation. In this approach, constitutive widest application to date. modification of the bait is achieved by its The rRas system, developed by fusion to a cognate modifying enzyme, Aronheim and colleagues (108), uses for example a kinase or an acetylase. the Ras pathway in yeast as a selection The fusion of substrate and enzyme mechanism. When localized at the results in efficient catalysis and ensures plasma membrane, the yeast Ras guanyl that the majority of the bait is posttransexchange factor cdc25 stimulates lationally modified. To demonstrate the guanyl nucleotide exchange on Ras and general usefulness of the approach, the promotes downstream signaling events authors identified yeast proteins that that ultimately lead to cell growth interact with acetylated histones and (108). A mutant yeast strain carrying proteins that specifically interact with the temperature-sensitive cdc25-2 the phosphorylated C-terminal tail of allele is able to grow at the permissive RNA polymerase II (102). Further temperature of 25°C, but fails to grow development of this system and its when shifted to 36°C (Figure 9A). The application to high-throughput interrRas system exploits this deficiency action screening may eventually lead to identify protein interactions at the to the creation of a posttranslational membrane. An integral membrane �

A

Vol. 40, No. 5 (2006)



protein (the bait) is expressed and targeted to the membrane (Figure 9A). An interacting protein (the prey) is expressed as a fusion to a constitutively active Ras mutant (mRas). Upon the interaction of bait and prey, mRas is translocated to the membrane, where it takes over the function of the defective cdc25 protein. Consequently, yeast harboring a protein interaction is able to grow at the nonpermissive temperature of 36°C (Figure 9B). Recently, the Ras system has been used to isolate interactors of small GTPases (109) and to develop a cellular screening assay for receptor tyrosine kinases (110). The MbYTH system, developed by our group (111), is an adaptation of the split-ubiquitin protein complementation assay (112). An integral membrane protein of interest (the bait) is fused to the C-terminal half of yeast ubiquitin (termed Cub), followed by a transcription factor composed of the Escherichia coli LexA protein and the VP16 transactivator derived from the herpes simplex virus. When no protein interaction takes place, the transcription factor is immobilized at the membrane and is unable to reach the nucleus (Figure 10A). An interacting protein (which can be an integral membrane protein or a cytosolic protein, the prey) is fused to the corresponding N-terminal half of ubiquitin (termed NubG). If bait and prey interact, the proximity of the Cub and NubG modules forces their association to form split-ubiquitin. The split-ubiquitin is recognized by ubiquitin-specific proteases (UBPs) present in the yeast cytosol. The UBPs cleave the polypeptide chain Cterminal to Cub, releasing the attached transcription factor, which then translocates to the nucleus and activates its cognate reporter genes (Figure 10B). The immediate advantage of the MbYTH system is the separation between protein interaction and the subsequent detection through a transcriptional output. For this reason, the system is suitable for assaying protein interactions that take place outside of the nucleus, for example at the plasma membrane (113). There is now a number of reports showing that the MbYTH system is capable of detecting interactions involving all types of integral membrane proteins BioTechniques 637

REVIEW and membrane-associated proteins (114–116), and it has also been used successfully to find novel interactors by cDNA library screening (117–119). In addition, the MbYTH system has recently been used on the large scale to identify interactions between S. cerevisiae integral membrane proteins. Among 705 selected integral membrane proteins, 1985 putative interactions involving 536 proteins were identified. The results identified potential new components of established biological processes and novel roles for integral membrane proteins of ascribed functions (120). As the detection of a protein interaction in the MbYTH system relies

A

A

X

Y

C

N L

HIS3 ADE2

B B Y

X

N

UB

P

C

L L HIS3 ADE2

Figure 10. The membrane yeast two-hybrid (MbYTH) system. (A) A bait is constructed by expressing an integral membrane protein or a membrane-associated protein X as a fusion to the C-terminal half of ubiquitin (C), followed by a transcription factor (L). Fusion of the tranFigure 10 scription factor to the integral membrane protein prevents its translocation to the nucleus and activation of the reporter genes. A prey is constructed by expressing an interacting protein Y as a fusion to the N-terminal half of ubiquitin (N). (B) If bait and prey interact, the N- and C-terminal halves of ubiquitin are forced into very close proximity and reassociate to form split-ubiquitin, which is recognized and cleaved by ubiquitin-specific proteases (UBP) in the cytosol. The cleavage liberates the transcription factor from the membrane, followed by its translocation to the nucleus and activation of reporter genes. 638 BioTechniques

on the same transcriptional mechanism as the yeast two-hybrid system, its adaptation to drug screening purposes should be easily possible by converting the abovementioned systems. The added advantage is that with the MbYTH system, full-length integral membrane proteins, which represent the majority of today’s drug targets, suddenly become accessible to screening. While it is well suited to integral membrane proteins, a drawback of the MbYTH system is that it cannot be used with cytosolic proteins as baits, since translocation of the soluble bait to the nucleus would activate the reporter genes in the absence of a protein interaction. This shortcoming can be prevented by attaching a cytosolic protein to the membrane by means of an additional membrane anchor (N. Möckli and D. Auerbach, unpublished). A different screening system based on the split-ubiquitin system has been developed by Johnsson and coworkers (121,122). Here, the protein attached to the C terminus of Cub consists of the yeast Ura3p protein. In the absence of a protein interaction, the Ura3p located on the cytosolic face of the membrane converts the compound 5FOA into a toxic metabolite, poisoning the yeast. Upon interactions with a NubG-fused prey and assembly of split-ubiquitin, the Ura3p is cleaved off, exposing a destabilizing N-terminal residue. Consequently, the destabilized Ura3p is rapidly degraded by the 26S proteasome, and yeast harboring the protein interaction is able to grow in the presence of 5FOA in the medium. When compared with the MbYTH system, the Ura3p-based system does not require attachment of the bait to the membrane, and it can therefore be used with cytosolic and even nuclear proteins (123). The option to screen nuclear proteins, such as transcription factors, presents another example of how newly developed technologies are overcoming disadvantages of the original yeast two-hybrid system. CONCLUSIONS AND FUTURE DIRECTIONS Functional genomics and proteomics are emerging subdisciplines of Systems Biology expected

to have a major impact on the future of biological and medical research. In order to fulfill this daunting task, the scientific community requires reliable model systems in which detailed studies can be performed accurately, rapidly, and reproducibly. Ever since Leland Hartwell and Paul Nurse set out to investigate the cell cycle using yeast as a model system, this simple fungus has seen a remarkable number of different applications, culminating in todays systems biology approaches that aim at describing cellular features such as gene expression, protein abundance, localization, and proteinprotein interactions at a genome-wide scale. Why has yeast been so hugely successful as a model organism? The answer probably lies in the unique combination of several features such as its genetic tractability, ease of manipulation, and the fact that many genes are highly conserved between yeast and human. Moreover, another important reason why yeast is a popular model organism lies in the unique yeast research community, which is well organized, open for exchange of data and materials, and has a demonstrated record in many joint achievements such as the sequencing of yeast genome, systematic knockout studies, synthetic genetic arrays, knock-in fusions of ORFs with GFP and TAP tags, and genome-wide micoarray studies. All these large-scale studies have yielded a wealth of data on yeast gene function and have provided important insights into how the approximately 6400 genes of yeast are making this unicellular eukaryote. Currently, most of this data are still deposited in single and largely unconnected databases. The future progress of functional genomics and proteomics using the budding yeast as a model system will involve refinement of current databases and the integration of data sets from different yeast-based systems biology studies. For example, data derived from largescale protein interaction studies can be combined with data from genetic interaction networks and transcriptome data to improve the general quality of the data set. To make the most out of the available data, it will be crucial to develop standards for recording data obtained by large-scale experiments Vol. 40, No. 5 (2006)

and to integrate existing and new data for better access. Another important future goal of the yeast research community will be to express and purify each yeast protein as well as to generate a collection of yeast strains in which each ORF is endogenously tagged with different short tags that can be recognized by commercially available antibodies. Very recently, the Grayhack and Snyder groups did an important step toward this goal by generating a versatile expression library that contains 5854 ORFs Cterminally tagged with 6×His and HA epitopes (124). In addition, they have demonstrated the effectiveness of this expression library to perform the first systematic survey of glycosylated proteins in yeast. Clearly, these new developments will be highly instructive in understanding the molecular function of the remaining approximately 2000 yeast ORFs whose function is not yet known. Toward this end, the yeast S. cerevisiae will continue to have a key role as an important model organism in the development of many current and future functional genomics and proteomics technologies. ACKNOWLEDGMENTS

We would like to thank Charlie Boone, Grant Brown, and Andres Lopes for critical reading of the manuscript and their helpful discussions. The research in the Stagljar group is supported by the Canadian government, the Union Bank of Switzerland, Novartis, Gebert Rüf Foundation, and the Swiss Cancer League (OCS-0131002-2003). COMPETING INTERESTS STATEMENT

The authors declare no competing interests. REFERENCES 1. Kumar, A. and M. Snyder. 2001. Emerging technologies in yeast genomics. Nat. Rev. Genet. 2:302-312. 2. Botstein, D. and G.R. Fink. 1988. Yeast: an experimental organism for modern biology. Science 240:1439-1443. Vol. 40, No. 5 (2006)

3. Botstein, D., S.A. Chervitz, and J.M. Cherry. 1997. Yeast as a model organism. Science 277:1259-1260. 4. Cherry, J.M., C. Ball, S. Weng, G. Juvik, R. Schmidt, C. Adler, B. Dunn, S. Dwight, et al. 1997. Genetic and physical maps of Saccharomyces cerevisiae. Nature 387:67-73. 5. Goffeau, A., B.G. Barrell, H. Bussey, R.W. Davis, B. Dujon, H. Feldmann, F. Galibert, J.D. Hoheisel, et al. 1996. Life with 6000 genes. Science 274:546-567. 6. Kumar, A., P.M. Harrison, K.H. Cheung, N. Lan, N. Echols, P. Bertone, P. Miller, M.B. Gerstein, et al. 2002. An integrated approach for finding overlooked genes in yeast. Nat. Biotechnol. 20:58-63. 7. Mewes, H.W., D. Frishman, U. Guldener, G. Mannhaupt, K. Mayer, M. Mokrejs, B. Morgenstern, M. Munsterkotter, et al. 2002. MIPS: a database for genomes and protein sequences. Nucleic Acids Res. 30:31-34. 8. Hudson, J.R., Jr., E.P. Dawson, K.L. Rushing, C.H. Jackson, D. Lockshon, D. Conover, C. Lanciault, J.R. Harris, et al. 1997. The complete set of predicted genes from Saccharomyces cerevisiae in a readily usable form. Genome Res. 7:1169-1173. 9. Wach, A., A. Brachat, R. Pohlmann, and P. Philippsen. 1994. New heterologous modules for classical or PCR-based gene disruptions in Saccharomyces cerevisiae. Yeast 10:1793-1808. 10. Mnaimneh, S., A.P. Davierwala, J. Haynes, J. Moffat, W.T. Peng, W. Zhang, X. Yang, J. Pootoolal, et al. 2004. Exploration of essential gene functions via titratable promoter alleles. Cell 118:31-44. 11. De Antoni, A. and D. Gallwitz. 2000. A novel multi-purpose cassette for repeated integrative epitope tagging of genes in Saccharomyces cerevisiae. Gene 246: 179-185. 12. Heitman, J., N.R. Movva, and M.N. Hall. 1991. Targets for cell cycle arrest by the immunosuppressant rapamycin in yeast. Science 253:905-909. 13. Hartwell, L.H. 2004. Yeast and cancer. Biosci. Rep. 24:523-544. 14. Giaever, G., A.M. Chu, L. Ni, C. Connelly, L. Riles, S. Veronneau, S. Dow, A. LucauDanila, et al. 2002. Functional profiling of the Saccharomyces cerevisiae genome. Nature 418:387-391. 15. Winzeler, E.A., D.D. Shoemaker, A. Astromoff, H. Liang, K. Anderson, B. Andre, R. Bangham, R. Benito, et al. 1999. Functional characterization of the S. cerevisiae genome by gene deletion and parallel analysis. Science 285:901-906. 16. Aouida, M., N. Page, A. Leduc, M. Peter, and D. Ramotar. 2004. A genome-wide screen in Saccharomyces cerevisiae reveals altered transport as a mechanism of resistance to the anticancer drug bleomycin. Cancer Res. 64:1102-1109. 17. Chang, M., M. Bellaoui, C. Boone, and G.W. Brown. 2002. A genome-wide screen for methyl methanesulfonate-sensitive mutants reveals genes required for S phase progression in the presence of DNA damage. Proc. Natl. Acad. Sci. USA 99:1693416939.

18. Huang, R.Y., M. Eddy, M. Vujcic, and D. Kowalski. 2005. Genome-wide screen identifies genes whose inactivation confer resistance to cisplatin in Saccharomyces cerevisiae. Cancer Res. 65:5890-5897. 19. Serrano, R., D. Bernal, E. Simon, and J. Arino. 2004. Copper and iron are the limiting factors for growth of the yeast Saccharomyces cerevisiae in an alkaline environment. J. Biol. Chem. 279:1969819704. 20. Zewail, A., M.W. Xie, Y. Xing, L. Lin, P.F. Zhang, W. Zou, J.P. Saxe, and J. Huang. 2003. Novel functions of the phosphatidylinositol metabolic pathway discovered by a chemical genomics screen with wortmannin. Proc. Natl. Acad. Sci. USA 100:33453350. 21. Parsons, A.B., R.L. Brost, H. Ding, Z. Li, C. Zhang, B. Sheikh, G.W. Brown, P.M. Kane, et al. 2004. Integration of chemicalgenetic and genetic interaction data links bioactive compounds to cellular target pathways. Nat. Biotechnol. 22:62-69. 22. Steinmetz, L.M., C. Scharfe, A.M. Deutschbauer, D. Mokranjac, Z.S. Herman, T. Jones, A.M. Chu, G. Giaever, et al. 2002. Systematic screen for human disease genes in yeast. Nat. Genet. 31:400404. 23. Perez, R.P. 1998. Cellular and molecular determinants of cisplatin resistance. Eur. J. Cancer 34:1535-1542. 24. Hartwell, L.H., P. Szankasi, C.J. Roberts, A.W. Murray, and S.H. Friend. 1997. Integrating genetic approaches into the discovery of anticancer drugs. Science 278:1064-1068. 25. Game, J.C., G.W. Birrell, J.A. Brown, T. Shibata, C. Baccari, A.M. Chu, M.S. Williamson, and J.M. Brown. 2003. Use of a genome-wide approach to identify new genes that control resistance of Saccharomyces cerevisiae to ionizing radiation. Radiat. Res. 160:14-24. 26. Birrell, G.W., G. Giaever, A.M. Chu, R.W. Davis, and J.M. Brown. 2001. A genomewide screen in Saccharomyces cerevisiae for genes affecting UV radiation sensitivity. Proc. Natl. Acad. Sci. USA 98:1260812613. 27. Lee, W., R.P. St Onge, M. Proctor, P. Flaherty, M.I. Jordan, A.P. Arkin, R.W. Davis, C. Nislow, et al. 2005. Genome-wide requirements for resistance to functionally distinct DNA-damaging agents. PLoS Genet. 1:e24. 28. Kaeberlein, M., R.W. Powers 3rd, K.K. Steffen, E.A. Westman, D. Hu, N. Dang, E.O. Kerr, K.T. Kirkland, et al. 2005. Regulation of yeast replicative life span by TOR and Sch9 in response to nutrients. Science 310:1193-1196. 29. Powers, R.W., 3rd, M. Kaeberlein, S.D. Caldwell, B.K. Kennedy, and S. Fields. 2006. Extension of chronological life span in yeast by decreased TOR pathway signaling. Genes Dev. 20:174-184. 30. Giaever, G., D.D. Shoemaker, T.W. Jones, H. Liang, E.A. Winzeler, A. Astromoff, and R.W. Davis. 1999. Genomic profiling of drug sensitivities via induced haploinsufficiency. Nat. Genet. 21:278-283. BioTechniques 639

REVIEW 31. Lum, P.Y., C.D. Armour, S.B. Stepaniants, G. Cavet, M.K. Wolf, J.S. Butler, J.C. Hinshaw, P. Garnier, et al. 2004. Discovering modes of action for therapeutic compounds using a genome-wide screen of yeast heterozygotes. Cell 116:121-137. 32. Giaever, G., P. Flaherty, J. Kumm, M. Proctor, C. Nislow, D.F. Jaramillo, A.M. Chu, M.I. Jordan, et al. 2004. Chemogenomic profiling: identifying the functional interactions of small molecules in yeast. Proc. Natl. Acad. Sci. USA 101:793798. 33. Kirsten, R., K. Nelson, D. Kirsten, and B. Heintz. 1998. Clinical pharmacokinetics of vasodilators. Part II. Clin. Pharmacokinet. 35:9-36. 34. Parker, W.B. and Y.C. Cheng. 1990. Metabolism and mechanism of action of 5fluorouracil. Pharmacol. Ther. 48:381-395. 35. Cherry, S., T. Doukas, S. Armknecht, S. Whelan, H. Wang, P. Sarnow, and N. Perrimon. 2005. Genome-wide RNAi screen reveals a specific sensitivity of IREScontaining RNA viruses to host translation inhibition. Genes Dev. 19:445-452. 36. Ashrafi, K., F.Y. Chang, J.L. Watts, A.G. Fraser, R.S. Kamath, J. Ahringer, and G. Ruvkun. 2003. Genome-wide RNAi analysis of Caenorhabditis elegans fat regulatory genes. Nature 421:268-272. 37. Kamath, R.S. and J. Ahringer. 2003. Genome-wide RNAi screening in Caenorhabditis elegans. Methods 30: 313-321. 38. Shi, Y. 2003. Mammalian RNAi for the masses. Trends Genet. 19:9-12. 39. Tong, A.H., M. Evangelista, A.B. Parsons, H. Xu, G.D. Bader, N. Page, M. Robinson, S. Raghibizadeh, et al. 2001. Systematic genetic analysis with ordered arrays of yeast deletion mutants. Science 294:2364-2368. 40. Tong, A.H., G. Lesage, G.D. Bader, H. Ding, H. Xu, X. Xin, J. Young, G.F. Berriz, et al. 2004. Global mapping of the yeast genetic interaction network. Science 303: 808-813. 41. Tong, A.H. and C. Boone. 2006. Synthetic genetic array analysis in Saccharomyces cerevisiae. Methods Mol. Biol. 313:171-192. 42. Bender, A. and J.R. Pringle. 1991. Use of a screen for synthetic lethal and multicopy suppressee mutants to identify two new genes involved in morphogenesis in Saccharomyces cerevisiae. Mol. Cell. Biol. 11:1295-1305. 43. Goehring, A.S., D.A. Mitchell, A.H. Tong, M.E. Keniry, C. Boone, and G.F. Sprague, Jr. 2003. Synthetic lethal analysis implicates Ste20p, a p21-activated potein kinase, in polarisome activation. Mol. Biol. Cell 14:1501-1516. 44. Measday, V., K. Baetz, J. Guzzo, K. Yuen, T. Kwok, B. Sheikh, H. Ding, R. Ueta, et al. 2005. Systematic yeast synthetic lethal and synthetic dosage lethal screens identify genes required for chromosome segregation. Proc. Natl. Acad. Sci. USA 102:1395613961. 45. Mayer, M.L., I. Pot, M. Chang, H. Xu, V. Aneliunas, T. Kwok, R. Newitt, R. Aebersold, et al. 2004. Identification of protein complexes required for efficient 640 BioTechniques

sister chromatid cohesion. Mol. Biol. Cell 15:1736-1745. 46. Bellaoui, M., M. Chang, J. Ou, H. Xu, C. Boone, and G.W. Brown. 2003. Elg1 forms an alternative RFC complex important for DNA replication and genome integrity. EMBO J. 22:4304-4313. 47. Chang, M., M. Bellaoui, C. Zhang, R. Desai, P. Morozov, L. Delgado-Cruzata, R. Rothstein, G.A. Freyer, et al. 2005. RMI1/ NCE4, a suppressor of genome instability, encodes a member of the RecQ helicase/ Topo III complex. EMBO J. 24:2024-2033. 48. Suter, B., A. Tong, M. Chang, L. Yu, G.W. Brown, C. Boone, and J. Rine. 2004. The origin recognition complex links replication, sister chromatid cohesion and transcriptional silencing in Saccharomyces cerevisiae. Genetics 167:579-591. 49. Lesage, G., J. Shapiro, C.A. Specht, A.M. Sdicu, P. Menard, S. Hussein, A.H. Tong, C. Boone, et al. 2005. An interactional network of genes involved in chitin synthesis in Saccharomyces cerevisiae. BMC Genet. 6:8. 50. Krogan, N.J., M. Kim, A. Tong, A. Golshani, G. Cagney, V. Canadien, D.P. Richards, B.K. Beattie, et al. 2003. Methylation of histone H3 by Set2 in Saccharomyces cerevisiae is linked to transcriptional elongation by RNA polymerase II. Mol. Cell. Biol. 23:4207-4218. 51. Davierwala, A.P., J. Haynes, Z. Li, R.L. Brost, M.D. Robinson, L. Yu, S. Mnaimneh, H. Ding, et al. 2005. The synthetic genetic interaction spectrum of essential genes. Nat. Genet. 37:1147-1152. 52. Schuldiner, M., S.R. Collins, N.J. Thompson, V. Denic, A. Bhamidipati, T. Punna, J. Ihmels, B. Andrews, et al. 2005. Exploration of the function and organization of the yeast early secretory pathway through an epistatic miniarray profile. Cell 123: 507-519. 53. Ooi, S.L., D.D. Shoemaker, and J.D. Boeke. 2003. DNA helicase gene interaction network defined using synthetic lethality analyzed by microarray. Nat. Genet. 35: 277-286. 54. Pan, X., D.S. Yuan, D. Xiang, X. Wang, S. Sookhai-Mahadeo, J.S. Bader, P. Hieter, F. Spencer, et al. 2004. A robust toolkit for functional profiling of the yeast genome. Mol. Cell 16:487-496. 55. Zhu, H., M. Bilgin, R. Bangham, D. Hall, A. Casamayor, P. Bertone, N. Lan, R. Jansen, et al. 2001. Global analysis of protein activities using proteome chips. Science 293:2101-2105. 56. Sopko, R., D. Huang, N. Preston, G. Chua, B. Papp, K. Kafadar, M. Synder, S.G. Oliver, et al. 2006. Mapping pathways and phenotypes by systemic gene overexpression. Mol. Cell 21:319-330. 57. Kamath, R.S., A.G. Fraser, Y. Dong, G. Poulin, R. Durbin, M. Gotta, A. Kanapin, N. Le Bot, et al. 2003. Systematic functional analysis of the Caenorhabditis elegans genome using RNAi. Nature 421:231-237. 58. Rual, J.F., J. Ceron, J. Koreth, T. Hao, A.S. Nicot, T. Hirozane-Kishikawa, J. Vandenhaute, S.H. Orkin, et al. 2004. Toward improving Caenorhabditis elegans

phenome mapping with an ORFeome-based RNAi library. Genome Res. 14:2162-2168. 59. Kumar, A., S. Agarwal, J.A. Heyman, S. Matson, M. Heidtman, S. Piccirillo, L. Umansky, A. Drawid, et al. 2002. Subcellular localization of the yeast proteome. Genes Dev. 16:707-719. 60. Huh, W.K., J.V. Falvo, L.C. Gerke, A.S. Carroll, R.W. Howson, J.S. Weissman, and E.K. O’Shea. 2003. Global analysis of protein localization in budding yeast. Nature 425:686-691. 61. Natter, K., P. Leitner, A. Faschinger, H. Wolinski, S. McCraith, S. Fields, and S.D. Kohlwein. 2005. The spatial organization of lipid synthesis in the yeast Saccharomyces cerevisiae derived from large scale green fluorescent protein tagging and high resolution microscopy. Mol. Cell. Proteomics 4:662-672. 62. Wigge, P.A., O.N. Jensen, S. Holmes, S. Soues, M. Mann, and J.V. Kilmartin. 1998. Analysis of the Saccharomyces spindle pole by matrix-assisted laser desorption/ ionization (MALDI) mass spectrometry. J. Cell Biol. 141:967-977. 63. Rout, M.P., J.D. Aitchison, A. Suprapto, K. Hjertaas, Y. Zhao, and B.T. Chait. 2000. The yeast nuclear pore complex: composition, architecture, and transport mechanism. J. Cell Biol. 148:635-651. 64. Ho, Y., A. Gruhler, A. Heilbut, G.D. Bader, L. Moore, S.L. Adams, A. Millar, P. Taylor, et al. 2002. Systematic identification of protein complexes in Saccharomyces cerevisiae by mass spectrometry. Nature 415:180-183. 65. Gavin, A.C., M. Bosche, R. Krause, P. Grandi, M. Marzioch, A. Bauer, J. Schultz, J.M. Rick, et al. 2002. Functional organization of the yeast proteome by systematic analysis of protein complexes. Nature 415:141-147. 66. Rigaut, G., A. Shevchenko, B. Rutz, M. Wilm, M. Mann, and B. Seraphin. 1999. A generic protein purification method for protein complex characterization and proteome exploration. Nat. Biotechnol. 17:1030-1032. 67. Puig, O., F. Caspary, G. Rigaut, B. Rutz, E. Bouveret, E. Bragado-Nilsson, M. Wilm, and B. Seraphin. 2001. The tandem affinity purification (TAP) method: a general procedure of protein complex purification. Methods 24:218-229. 68. Gavin, A.C., P. Aloy, P. Grandi, R. Krause, M. Boesche, M. Marzioch, C. Rau, L.J. Jensen, et al. 2006. Proteome survey reveals modularity of the yeast cell machinery. Nature Jan 22; [Epub ahead of print]. 69. Knuesel, M., Y. Wan, Z. Xiao, E. Holinger, N. Lowe, W. Wang, and X. Liu. 2003. Identification of novel protein-protein interactions using a versatile Mammalian tandem affinity purification expression system. Mol. Cell. Proteomics 2:1225-1233. 70. Ghaemmaghami, S., W.K. Huh, K. Bower, R.W. Howson, A. Belle, N. Dephoure, E.K. O’Shea, and J.S. Weissman. 2003. Global analysis of protein expression in yeast. Nature 425:737-741. 71. Krogan, N.J., G. Cagney, H. Yu, G. Zhong, X. Guo, A. Ignatchenko, J. Li, S. Pu, et Vol. 40, No. 5 (2006)

al. 2006. Global landscape of protein complexes in the yeast Saccharomyces cerevisiae. Nature 2006 March 22; [Epub ahead of print]. 72. Angenendt, P. 2005. Progress in protein and antibody microarray technology. Drug Discov. Today 10:503-511. 73. Zhu, H. and M. Snyder. 2003. Protein chip technology. Curr. Opin. Chem. Biol. 7:5563. 74. MacBeath, G. and S.L. Schreiber. 2000. Printing proteins as microarrays for highthroughput function determination. Science 289:1760-1763. 75. Ptacek, J., G. Devgan, G. Michaud, H. Zhu, X. Zhu, J. Fasolo, H. Guo, G. Jona, et al. 2005. Global analysis of protein phosphorylation in yeast. Nature 438:679-684. 76. Michaud, G.A., M. Salcius, F. Zhou, R. Bangham, J. Bonin, H. Guo, M. Snyder, P.F. Predki, et al. 2003. Analyzing antibody specificity with whole proteome microarrays. Nat. Biotechnol. 21:1509-1512. 77. Hall, D.A., H. Zhu, X. Zhu, T. Royce, M. Gerstein, and M. Snyder. 2004. Regulation of gene expression by a metabolic enzyme. Science 306:482-484. 78. Huang, J., H. Zhu, S.J. Haggarty, D.R. Spring, H. Hwang, F. Jin, M. Snyder, and S.L. Schreiber. 2004. Finding new components of the target of rapamycin (TOR) signaling network through chemical genetics and proteome chips. Proc. Natl. Acad. Sci. USA 101:16594-16599. 79. Schmelzle, T. and M.N. Hall. 2000. TOR, a central controller of cell growth. Cell 103:253-262. 80. Fields, S. and O. Song. 1989. A novel genetic system to detect protein-protein interactions. Nature 340:245-246. 81. Van Criekinge, W. and R. Beyaert. 1999. Yeast two-hybrid: state of the art. Biol. Proced. Online 2:1-38. 82. Fashena, S.J., I. Serebriiskii, and E.A. Golemis. 2000. The continued evolution of two-hybrid screening approaches in yeast: how to outwit different preys with different baits. Gene 250:1-14. 83. Xenarios, I., D.W. Rice, L. Salwinski, M.K. Baron, E.M. Marcotte, and D. Eisenberg. 2000. DIP: the database of interacting proteins. Nucleic Acids Res. 28: 289-291. 84. Uetz, P., L. Giot, G. Cagney, T.A. Mansfield, R.S. Judson, J.R. Knight, D. Lockshon, V. Narayan, et al. 2000. A comprehensive analysis of protein-protein interactions in Saccharomyces cerevisiae. Nature 403:623-627. 85. McCraith, S., T. Holtzman, B. Moss, and S. Fields. 2000. Genome-wide analysis of vaccinia virus protein-protein interactions. Proc. Natl. Acad. Sci. USA 97:4879-4884. 86. Flajolet, M., G. Rotondo, L. Daviet, F. Bergametti, G. Inchauspe, P. Tiollais, C. Transy, and P. Legrain. 2000. A genomic approach of the hepatitis C virus generates a protein interaction map. Gene 242:369-379. 87. Rain, J.C., L. Selig, H. De Reuse, V. Battaglia, C. Reverdy, S. Simon, G. Lenzen, F. Petel, et al. 2001. The protein-

Vol. 40, No. 5 (2006)

protein interaction map of Helicobacter pylori. Nature 409:211-215. 88. Ito, T., T. Chiba, R. Ozawa, M. Yoshida, M. Hattori, and Y. Sakaki. 2001. A comprehensive two-hybrid analysis to explore the yeast protein interactome. Proc. Natl. Acad. Sci. USA 98:4569-4574. 89. Giot, L., J.S. Bader, C. Brouwer, A. Chaudhuri, B. Kuang, Y. Li, Y.L. Hao, C.E. Ooi, et al. 2003. A protein interaction map of Drosophila melanogaster. Science 302:1727-1736. 90. Li, S., C.M. Armstrong, N. Bertin, H. Ge, S. Milstein, M. Boxem, P.O. Vidalain, J.D. Han, et al. 2004. A map of the interactome network of the metazoan C. elegans. Science 303:540-543. 91. Rual, J.F., K. Venkatesan, T. Hao, T. Hirozane-Kishikawa, A. Dricot, N. Li, G.F. Berriz, F.D. Gibbons, et al. 2005. Towards a proteome-scale map of the human protein-protein interaction network. Nature 437:1173-1178. 92. Stelzl, U., U. Worm, M. Lalowski, C. Haenig, F.H. Brembeck, H. Goehler, M. Stroedicke, M. Zenkner, et al. 2005. A human protein-protein interaction network: a resource for annotating the proteome. Cell 122:957-968. 93. Stanyon, C.A., G. Liu, B.A. Mangiola, N. Patel, L. Giot, B. Kuang, H. Zhang, J. Zhong, et al. 2004. A Drosophila proteininteraction map centered on cell-cycle regulators. Genome Biol. 5:R96. 94. LaCount, D.J., M. Vignali, R. Chettier, A. Phansalker, R. Bell, J.R. Hesselberth, L.W. Schoenfeld, I. Ota, et al. 2005. A protein interaction network of the malaria parasite Plasmodium falciparum. Nature 438:103-107. 95. Zhong, J., H. Zhang, C.A. Stanyon, G. Tromp, and R.L. Finley, Jr. 2003. A strategy for constructing large protein interaction maps using the yeast two-hybrid system: regulated expression arrays and two-phase mating. Genome Res. 13:2691-2699. 96. Baker, K., D. Sengupta, G. SalazarJimenez, and V.W. Cornish. 2003. An optimized dexamethasone-methotrexate yeast 3-hybrid system for high-throughput screening of small molecule-protein interactions. Anal. Biochem. 315:134-137. 97. Henthorn, D.C., A.A. Jaxa-Chamiec, and E. Meldrum. 2002. A GAL4-based yeast three-hybrid system for the identification of small molecule-target protein interactions. Biochem. Pharmacol. 63:1619-1628. 98. Vidal, M., R.K. Brachmann, A. Fattaey, E. Harlow, and J.D. Boeke. 1996. Reverse two-hybrid and one-hybrid systems to detect dissociation of protein-protein and DNAprotein interactions. Proc. Natl. Acad. Sci. USA 93:10315-10320. 99. Huang, J. and S.L. Schreiber. 1997. A yeast genetic system for selecting small molecule inhibitors of protein-protein interactions in nanodroplets. Proc. Natl. Acad. Sci. USA 94:13396-13401. 100. Kato-Stankiewicz, J., I. Hakimi, G. Zhi, J. Zhang, I. Serebriiskii, L. Guo, H. Edamatsu, H. Koide, et al. 2002. Inhibitors of Ras/Raf-1 interaction identified by two-

hybrid screening revert Ras-dependent transformation phenotypes in human cancer cells. Proc. Natl. Acad. Sci. USA 99:1439814403. 101. Auerbach, D., S. Thaminy, M.O. Hottiger, and I. Stagljar. 2002. The post-genomic era of interactive proteomics: facts and perspectives. Proteomics 2:611-623. 102. Guo, D., T.R. Hazbun, X.J. Xu, S.L. Ng, S. Fields, and M.H. Kuo. 2004. A tethered catalysis, two-hybrid system to identify protein-protein interactions requiring posttranslational modifications. Nat. Biotechnol. 22:888-892. 103. Aronheim, A. 1997. Improved efficiency sos recruitment system: expression of the mammalian GAP reduces isolation of Ras GTPase false positives. Nucleic Acids Res. 25:3373-3374. 104. Marsolier, M.-C., M.-N. Prioleau, and A. Sentenac. 1997. A RNA polymerase III-based two-hybrid system to study RNA polymerase II transcriptional regulators. J. Mol. Biol. 268:243-249. 105. Ehrhard, K.N., J.J. Jacoby, X.Y. Fu, R. Jahn, and H.G. Dohlman. 2000. Use of G-protein fusions to monitor integral membrane protein-protein interactions in yeast. Nat. Biotechnol. 18:1075-1079. 106. Urech, D.M., P. Lichtlen, and A. Barberis. 2003. Cell growth selection system to detect extracellular and transmembrane protein interactions. Biochim. Biophys. Acta 1622:117-127. 107. Hubsman, M., G. Yudkovsky, and A. Aronheim. 2001. A novel approach for the identification of protein-protein interaction with integral membrane proteins. Nucleic Acids Res. 29:E18. 108. Aronheim, A., D. Engelberg, N. Li, N. al-Alawi, J. Schlessinger, and M. Karin. 1994. Membrane targeting of the nucleotide exchange factor Sos is sufficient for activating the Ras signaling pathway. Cell 78:949961. 109. Frankel, P., A. Aronheim, E. Kavanagh, M.S. Balda, K. Matter, T.D. Bunney, and C.J. Marshall. 2005. RalA interacts with ZONAB in a cell density-dependent manner and regulates its transcriptional activity. EMBO J. 24:54-62. 110. Gunde, T. and A. Barberis. 2005. Yeast growth selection system for detecting activity and inhibition of dimerization-dependent receptor tyrosine kinase. BioTechniques 39:541-549. 111. Stagljar, I., C. Korostensky, N. Johnsson, and S. te Heesen. 1998. A genetic system based on split-ubiquitin for the analysis of interactions between membrane proteins in vivo. Proc. Natl. Acad. Sci. USA 95:51875192. 112. Johnsson, N. and A. Varshavsky. 1994. Split ubiquitin as a sensor of protein interactions in vivo. Proc. Natl. Acad. Sci. USA 91:10340-10344. 113. Iyer, K., L. Bürkle, D. Auerbach, S. Thaminy, M. Dinkel, K. Engels, and I. Stagljar. 2005. Utilizing the yeast split ubiquitin membrane yeast two-hybrid system (MbYTH) to identify interacting proteins of membrane proteins. Sci. STKE 275:p13. BioTechniques 641

REVIEW 114. Obrdlik, P., M. El-Bakkoury, T. Hamacher, C. Cappellaro, C. Vilarino, C. Fleischer, H. Ellerbrok, R. Kamuzinzi, et al. 2004. K+ channel interactions detected by a genetic system optimized for systematic studies of membrane protein interactions. Proc. Natl. Acad. Sci. USA 101:1224212247. 115. Pandey, S. and S.M. Assmann. 2004. The Arabidopsis putative G protein-coupled receptor GCR1 interacts with the G protein alpha subunit GPA1 and regulates abscisic acid signaling. Plant Cell 16:1616-1632. 116. Yan, A., E. Wu, and W.J. Lennarz. 2005. Studies of yeast oligosaccharyl transferase subunits using the split-ubiquitin system: topological features and in vivo interactions. Proc. Natl. Acad. Sci. USA 102:7121-7126. 117. Thaminy, S., D. Auerbach, A. Arnoldo, and I. Stagljar. 2003. Identification of novel ErbB3-interacting factors using the split-ubiquitin membrane yeast two-hybrid system. Genome Res. 13:1744-1753. 118. Wang, B., J. Pelletier, M.J. Massaad, A. Herscovics, and G.C. Shore. 2004. The yeast split-ubiquitin membrane protein twohybrid screen identifies BAP31 as a regulator of the turnover of endoplasmic reticulumassociated protein tyrosine phosphatase-like B. Mol. Cell. Biol. 24:2767-2778. 119. Matsuda, S., L. Giliberto, Y. Matsuda, P. Davies, E. McGowan, F. Pickford, J. Ghiso, B. Frangione, et al. 2005. The familial dementia BRI2 gene binds the Alzheimer gene amyloid-beta precursor protein and inhibits amyloid-beta production. J. Biol. Chem. 280:28912-28916. 120. Miller, J.P., R.S. Lo, A. Ben-Hur, C. Desmarais, I. Stagljar, W.S. Noble, and S. Fields. 2005. Large-scale identification of yeast integral membrane protein interactions. Proc. Natl. Acad. Sci. USA 102:12123-12128. 121. Wittke, S., N. Lewke, S. Muller, and N. Johnsson. 1999. Probing the molecular environment of membrane proteins in vivo. Mol. Biol. Cell 10:2519-2530. 122. Dunnwald, M., A. Varshavsky, and N. Johnsson. 1999. Detection of transient in vivo interactions between substrate and transporter during protein translocation into the endoplasmic reticulum. Mol. Biol. Cell 10:329-344. 123. Laser, H., C. Bongards, J. Schuller, S. Heck, N. Johnsson, and N. Lehming. 2000. A new screen for protein interactions reveals that the Saccharomyces cerevisiae high mobility group proteins Nhp6A/B are involved in the regulation of the GAL1 promoter. Proc. Natl. Acad. Sci. USA 97:1373213737. 124. Gelperin, D.M., M.A. White, M.L. Wilkinson, Y. Kon, L.A. Kung, K.J. Wise, N. Lopez-Hoyo, L. Jiang, et al. 2005. Biochemical and genetic analysis of the yeast proteome with a movable ORF collection. Genes Dev. 19:2816-2826.

Donnelly Centre for Cellular and Biomolecular Research University of Toronto 160 College Street Toronto, ON, M5S 3E1, Canada e-mail: [email protected]

To purchase reprints of this article, contact [email protected]

Address correspondence to: Igor Stagljar Department of Biochemistry and Department of Medical Genetics & Microbiology 642 BioTechniques

Vol. 40, No. 5 (2006)

Yeast-based functional genomics and proteomics ... - Semantic Scholar

... the cellular milieu. 1University of Toronto, Toronto, ON, Canada and 2Dualsystems Biotech Inc., Zurich, Switzerland ...... 160 College Street. Toronto, ON, M5S ...

527KB Sizes 3 Downloads 199 Views

Recommend Documents

Yeast-based functional genomics and proteomics ... - Semantic Scholar
... cellular milieu. 1University of Toronto, Toronto, ON, Canada and 2Dualsystems Biotech Inc., Zurich, Switzerland ...... AD. Y. HIS3. HIS3. HIS3. Figure 7. Small molecule yeast two-hybrid screening. .... co-express the modifying enzyme along.

Yeast-based functional genomics and proteomics ... - BioTechniques
Oliver, et al. 2006. Mapping pathways and phenotypes by systemic gene overexpres- sion. Mol. Cell 21:319-330. 57. Kamath, R.S., A.G. Fraser, Y. Dong, G. Poulin, R. Durbin, M. Gotta, A. Kanapin,. N. Le Bot, et al. 2003. Systematic function- al analysi

Yeast-based functional genomics and proteomics ... - BioTechniques
org), the Yeast Protein Database. (YPD ... a big advantage of yeast compared with ... able in the drive toward a comprehensive understanding of protein structure and function in the cellular milieu. ... these genes from the data sets revealed.

Yeast-based functional genomics and proteomics ... - BioTechniques
and the Yeast Resource Center (depts. washington.edu/~yeastrc). ...... identified with at least one partner (68). One major concern is ...... of this article, contact.

Integration and Visualisation of Genomics and Proteomics Data - GitHub
This enables co-visualization of genomics, transcriptomics, and proteomics data using the Integrative ... The software is available via the GitHub code repository:.

pdf-08108\genomics-proteomics-and-metabolomics-in ...
... apps below to open or edit this item. pdf-08108\genomics-proteomics-and-metabolomics-in-nut ... uticals-and-functional-foods-from-wiley-blackwell.pdf.

Genomics and Proteomics of Type 2 Diabetes in ... - MyScienceWork
development of disease not only shows exponential growth but also ... associated with type 2 diabetes in Asian Indian subjects and the XA ... Among various applications, clinical .... of view, the cost of applyingthese technologies and the.