Improving the yeast two-hybrid system with permutated fusions proteins: the Varicella Zoster Virus interactome
© Stellberger et al; licensee BioMed Central Ltd. 2010
Received: 19 December 2009
Accepted: 15 February 2010
Published: 15 February 2010
Yeast two-hybrid (Y2H) screens have been among the most powerful methods to detect and analyze protein-protein interactions. However, they suffer from a significant degree of false negatives, i.e. true interactions that are not detected, and to a certain degree from false positives, i.e. interactions that appear to take place only in the context of the Y2H assay. While the fraction of false positives remains difficult to estimate, the fraction of false negatives in typical Y2H screens is on the order of 70-90%. Here we present novel Y2H vectors that significantly decrease the number of false negatives and help to mitigate the false positive problem.
We have constructed two new vectors (pGBKCg and pGADCg) that allow us to make both C-terminal fusion proteins of DNA-binding and activation domains. Both vectors can be combined with existing vectors for N-terminal fusions and thus allow four different bait-prey combinations: NN, CC, NC, and CN. We have tested all ~4,900 pairwise combinations of the 70 Varicella-Zoster-Virus (VZV) proteins for interactions, using all possible combinations. About ~20,000 individual Y2H tests resulted in 182 NN, 89 NC, 149 CN, and 144 CC interactions. Overlap between screens ranged from 17% (NC-CN) to 43% (CN-CC). Performing four screens (i.e. permutations) instead of one resulted in about twice as many interactions and thus much fewer false negatives. In addition, interactions that are found in multiple combinations confirm each other and thus provide a quality score. This study is the first systematic analysis of such N- and C-terminal Y2H vectors.
Permutations of C- and N-terminal Y2H vectors dramatically increase the coverage of interactome studies and thus significantly reduce the number of false negatives. We suggest that future interaction screens should use such vector combinations on a routine basis, not the least because they provide a built-in quality score for Y2H interactions that can provide a measure of reproducibility without additional assays.
The yeast two-hybrid (Y2H) system has been among the most powerful methods to identify protein-protein interactions. However, it has also been criticized for generating large numbers of false positive and false negative data. While false positives can be minimized by including controls, retesting by independent methods, and bioinformatic filtering, false negatives pose a much bigger problem. In fact, many interactions must go undetected in two-hybrid screens because of the sterical constraints the system involves: the two fusion proteins must interact with each other, their interaction interfaces must be exposed, and all components must be oriented on the DNA in a way that the activation domain of the prey can productively interact with the transcriptional machinery.
On top of sterical constraints, small variations of the Y2H system can lead to large differences in resulting interactions . As no other large-scale screen has used multiple Y2H systems, not much attention has been paid to this phenomenon. Here we investigate such multi-system screens to characterize the interactome of Varicella-Zoster Virus (VZV). While many smaller viruses and their interactions have been studied, larger viruses have been analyzed only recently [2, 3]. The five herpesviruses studied by Fossum et al. (i.e. HSV1, EBV, mCMV, VZV, KSHV) have been the only herpesviruses studied systematically for pairwise intraviral interactions . However, we also know that the few studies that have been done resulted in markedly different interactions. For example, two systematic screens of protein-protein interactions among proteins of Kaposi-Sarcoma Herpesvirus (KSHV) yielded very little overlap [5, 6]. Clearly, slight variations of vectors, strains, or assay conditions can strongly affect the resulting interactions, even when identical proteins are used.
Interactions resulting from different vectors
The interactions detected also depended strongly on the selection pressure: typically all screens were initially carried out without 3-aminotriazole (3AT), a competitive inhibitor of imidazoleglycerolphosphate (IGP) dehydratase, i.e. the His3 reporter enzyme used in our assays. If baits turned out to be autoactivating under these conditions, we raised the 3AT concentrations up to 50 mM in steps of 1, 3, 10, 25, and 50 mM. ORF19 was an activator at 0, 1, and 3 mM and clear results were only obtained at 10 or 25 mM (Figure 3). Most interactions disappeared at 25 mM (Figure 3B). In general, the results reported here were obtained at a 3AT concentration that clearly differentiated between signal and noise. Note that ORF19 was an activator as an N-terminal DBD-fusion but not as a C-terminal fusion (background in Figure 3A). However, this was not generally true: in 20 cases the N-terminal bait fusion autoactivated while 21 of the C-terminal baits did so. In 7 cases both fusions were autoactivators at 3 mM 3AT or higher concentrations but only one (ORF46) required more than 3 mM in both cases and was thus not interpretable (data not shown).
Overlap between vector combinations
Overlap between screens
Reduction of false negatives
Similarly, we asked how many of 166 Y2H interactions found among core proteins of HSV1, mCMV, EBV, or KSHV are found in at least one of our 4 vector combinations (core proteins are those proteins that are conserved in all 5 viruses). Again, given that not all interactions are conserved, we do not expect to find all of these interactions. If single screens are performed (i.e. with only one of the 4 vector pairs) we find 14% of these core interactions on average (Figure 5B, Additional file 1: Table S3E). However, if we use all four vector combinations, this fractions more than doubles to 31% or 51 interactions. This clearly demonstrates that the use of multiple fusion proteins can significantly reduce the number of false negatives in Y2H screens.
Reduction of false positives
It is almost impossible to prove the existence of false positives, especially for viral proteins that are known to be expressed in the same cell. We simply cannot know for sure that interactions detected in a Y2H assay do not happen in a human host cell. Thus we provide only circumstantial evidence that interactions found in multiple combinations are of higher quality than interactions found in a single screen. We looked at interactions that we found in one, two, three, or four combinations and then determined the fraction of these subsets that have been confirmed by either literature curated interologs, Y2H interologs, or independent Y2H assays. The latter include cases in which an interaction is found with different protein fragments, e.g. the interaction shown in Figure 3 involves ORF18 as well as ORF18C which is a C-terminal fragment of ORF18. We believe that interactions found with multiple protein fragments support this interaction (as long as the fragments are not exclusive as in non-overlapping N- and C-terminal fragments). Figure 5C shows that the number of combinations that an interaction was found in is directly correlated with the level of confirmation. That is, the more combinations an interaction is found with, the more likely it is to be found in our literature-curated gold standard dataset, among a set of interologs, or among additional constructs. In other words: if an interaction is found in two or more combinations it is less likely to be a false positive.
Which vector combination is best?
We wondered whether any of the four permutations was superior to the other ones. Surprisingly, this does not seem to be the case. The fraction of validated interactions (based on interologs and validating alternative constructs) in all cases ranges from 45% to 50% (Additional file 1: Table S3C).
The main problem of Y2H screens is the high rate of false negatives [11, 12] and new vector systems as the one shown here may alleviate that problem. We have shown recently that alternative N-terminal vector systems can reduce the number of false positives and this study extends this finding to C-terminal fusions and combinations with N-terminal vectors.
Expression levels of bait and prey proteins
Results of Y2H assays may be affected by the expression levels of bait and prey proteins, e.g. if one or both are inefficiently translated or if the proteins themselves are unstable. We have not tested the expression levels of the various N- and C-terminal fusion proteins used here but we believe that their expression levels are similar or at least sufficient for several reasons. Most importantly, we get similar numbers of interactions with N- and C-terminal fusion proteins, indicating that they both form functional (or at least interacting) proteins, even if the NC combination yielded a significantly lower number of interations than the CN combination.
Two-hybrid interactions have often been considered as unreliable, generating many false positive and false negative results . Often false positives are suspected because many interactions are implausible and they lack independent confirmation. However, while additional assays such as co-immunoprecipitation can corroborate Y2H interactions, they are often time-consuming, expensive, and require additional reagents and protocols. In addition, a large fraction of bona fide interactions may be suitable for one particular assay but not for another and thus confirmatory experiments may still miss up to 80% of all true interactions . Our results show that permutating fusion proteins in Y2H screens may mitigate these problems without requiring additional assays.
Further permutations of the Y2H system
Obviously many more variations of the Y2H system are possible in addition to changing the location and nature of fusion tags, e.g. copy number, yeast strains, reporters etc. . Other Y2H systems including commercial vectors may benefit from the permutations described and they should be used routinely in large-scale screens.
The parental vectors and the two new C-terminal vectors described here are identical in most ways, including their origins of replication (and thus their copy number), their promoters (and thus protein expression levels), as well as the yeast strains in which they were expressed. In addition, the experimental conditions of our Y2H assays were identical (except for varying 3AT concentrations). Hence, we conclude that differences in results must result from structural constraints imposed by the location of the DBD and AD fusion tags and the associated sequences.
Novel VZV interactions: ORF10-ORF57
Interactions that are found in multiple permutations but have not been reported in the literature are the most interesting ones for follow-up studies. For space reasons we cannot discuss all interactions that we consider as well substantiated by multiple Y2H assays. One intriguing example is the interaction between ORF57 and ORF10 which we found in all four permutations. ORF10 protein is the homologue of the herpes simplex virus type 1 (HSV-1) VP16 protein. Both proteins have been shown to be structural components of the virion tegument, but also appear to activate viral immediate-early (IE) protein promoters . In contrast to HSV-1, where ORF10 is essential, VZV ORF10 is dispensable for viral replication in vitro . However, a more detailed analysis of VZV ORF10 deletion mutants in SCIDhu skin xenografts revealed that ORF10 is a virulence factor for the pathogenesis of VZV in skin [16, 17]. VZV ORF10 mutants are characterized by decreased viral titers and decreased cutaneous lesions within skin xenografts. Electron microscopy (EM) pictures showed that VZV-infected epidermal cells had significantly fewer DNA containing nucleocapsids and extensive aggregates of intracytoplasmic viral particles .
Little is known about the function of the VZV ORF57 protein. This protein has been shown to localize to the cytoplasm of infected cells and to be dispensable for the replication of VZV in vitro . However, UL3.5, the homologue of VZV ORF57 in pseudorabies virus (PRV), has been demonstrated to be essential for virus replication in vitro, playing an important role during viral egress. EM pictures showed that cells infected with the PRV UL3.5 deletion mutants exhibited accumulated intracytoplasmic viral capsids [19, 20]. Deletion mutants of VZV ORF10 and the UL3.5 homologue of VZV ORF57 in PRV exhibit identical phenotypes, i.e. the intracytoplasmic aggregation of viral particles, and thus suggest that both proteins play an essential role in virus egress. However, their precise molecular role in this process remains to be clarified.
Our data contains a total of 15 pairs that are found in three or four permutations but have never reported in the literature (Additional file 1: Table S1). They should be excellent candidates for further studies of VZV molecular and structural biology.
Recommendations for future interactome screens
As a direct result from our data we recommend that future screens should be done with a combinatin of vectors. For example, mixtures of N- and C-terminal bait plasmids could be mated with mixed N- and C-terminal prey libraries (or arrays). As with conventional screens bait constructs would have to be tested for auto-activation but this remains straight-forward. In order to improve mating efficiency the four combinations (NN, NC, CN, CC) could be mated separately and then mixed for further selection of positives. However, for less complex libraries (as those of virus genomes) this should not be necessary. Given that other factors in addition to N- or C-terminal fusions play a role in Y2H screens , additional vectors could be added.
In summary, using permutations of fusion proteins has the potential to dramatically reduce false negatives in yeast two-hybrid screens while providing a simple way to rank interactions qualitatively. This should significantly improve future large-scale screens as well as downstream analysis of selected interactions.
Literature-curated interaction data and interologs
Literature-curated interactions were derived from : First, a few VZV interactions from small-scale studies are considered to be highly reliable as they have been verified by additional experiments. Second, small-scale interactions from other herpesviruses with homologs in VZV are here considered as "literature-curated" as well. Third, we used large-scale Y2H data from  and other studies. If these interactions (in any herpesvirus different from VZV) had homologous pairs in VZV, they were considered as "interologs". For example, the KSHV interaction ORF60-ORF61 has a homologous interacting pair in VZV, ORF18-ORF19 which is thus an interolog of ORF60-ORF61.
Vector construction and ORF cloning
Here we describe a new set of vectors that are based on the well-established pGBKT7 and pGADT7 vectors (Clontech) and their Gateway derivatives pGBKT7g and pGADT7g  (Figure 1). We modified the latter two vectors by inserting the Gateway RfB cassette (for reading frame B) in a way that the DBD and AD can be fused to the C-termini of cloned proteins as opposed to the usual N-termini and named these vectors pGBKCg and pGADCg, respectively (Figure 2). Next, we cloned all full-length 69 open reading frames (ORFs) as well as 18 protein fragments of Varicella-Zoster-Virus (VZV) into pGADT7g, pGBKT7g, pGADCg, and pGBKCg (Additional file 1: Table S4).
Yeast two-hybrid assays and quality scoring
We systematically tested all N- and C-terminal baits against all N- and C-terminal VZV preys using standard matrix-based Y2H protocols as described previously [21, 22]. The quality of interactions was evaluated by two different strategies: first, we compared our interactions to a list of known herpesviral interactions. Because not many interactions have been detected in VZV (except in our own studies [4, 5], we used a list of herpesviral interactions curated from small-scale studies  as a gold-standard dataset. We considered a VZV interaction as "LC-verified" if it had at least one interolog in this literature-curated (LC) dataset. Second, we compared our VZV dataset to a large set of herpesviral Y2H interactions , comprising interactions from KSHV, VZV, EBV, mCMV, and HSV1. We considered a VZV interaction as "interolog-verified" if it had at least one interolog in one of the four non-VZV viruses (as VZV interactions would not be interologs). This logic is based on the idea that interactions are conserved and that multiple interologs support an interaction [23–25]. All interactions, interologs, and literature-curated interactions are available as Additional file 1: Tables S1-3 and from .
S.V. Rajagopala made useful comments on the manuscript. This project was funded by the Landesstiftung Baden-Württemberg (Germany) and NIH grant RO1GM79710 to PU. Financial support by the Deutsche Forschungsgemeinschaft (BA 2035/3-1) to AB and JH and by the Bayerisches Staatsministerium fuer Wissenschaft, Kultur und Kunst (Bayerisches Genomforschungsnetzwerk) to JH is gratefully acknowledged. The funding agencies had no other role in this project.
- Rajagopala SV, Hughes KT, Uetz P: Benchmarking yeast two-hybrid systems using the interactions of bacterial motility proteins. Proteomics 2009, 9: 5296–5302. 10.1002/pmic.200900282PubMed CentralPubMedView Article
- Uetz P, Rajagopala SV, Dong YA, Haas J: From ORFeomes to protein interaction maps in viruses. Genome Res 2004, 14: 2029–2033. 10.1101/gr.2583304PubMedView Article
- Mendez-Rios J, Uetz P: Global approaches to study protein-protein interactions among viruses and hosts. Future Microbiology 2010, in press.
- Fossum E, Friedel CC, Rajagopala SV, Titz B, Baiker A, Schmidt T, Kraus T, Stellberger T, Rutenberg C, Suthram S, Bandyopadhyay S, Rose D, Brunn Av, Uhlmann M, Zeretzke C, Dong YA, Boulet H, Koegl M, Bailer SM, Koszinowski U, Ideker T, Uetz P, Zimmer R, Haas J: Evolutionarily conserved herpesviral protein interaction networks. PloS Pathogens 2009, 5: e1000570. 10.1371/journal.ppat.1000570PubMed CentralPubMedView Article
- Uetz P, Dong YA, Zeretzke C, Atzler C, Baiker A, Berger B, Rajagopala SV, Roupelieva M, Rose D, Fossum E, Haas J: Herpesviral protein networks and their interaction with the human proteome. Science 2006, 311: 239–242. 10.1126/science.1116804PubMedView Article
- Rozen R, Sathish N, Li Y, Yuan Y: Virion-wide protein interactions of Kaposi's sarcoma-associated herpesvirus. J Virol 2008, 82: 4742–4750. 10.1128/JVI.02745-07PubMed CentralPubMedView Article
- Brown MA, MacGillivray RT: Vectors for expressing proteins at the amino-terminus of an activation domain for use in the yeast two-hybrid system. Anal Biochem 1997, 247: 451–452. 10.1006/abio.1997.2118PubMedView Article
- James P: Yeast Two-Hybrid Vectors and Strains. In Two-Hybrid Systems Methods and Protocols. Volume 177. Edited by: MacDonald PN. Totowa, New Jersey: Humana Press; 2001. [Walker JM (Series Editor): Methods in Molecular Biology]
- Millson SH, Truman AW, Piper PW: Vectors for N- or C-terminal positioning of the yeast Gal4p DNA binding or activator domains. Biotechniques 2003, 35: 60–64.PubMed
- Béranger F, Aresta S, Gunzburg Jd, Camonis J: Getting more from the two-hybrid system: N-terminal fusions to LexA are efficient and sensitive baits for two-hybrid studies. Nucleic Acids Research 1997, 25: 2035–2036. 10.1093/nar/25.10.2035PubMed CentralPubMedView Article
- Rajagopala SV, Titz B, Goll J, Parrish JR, Wohlbold K, McKevitt MT, Palzkill T, Mori H, Finley RL Jr, Uetz P: The protein network of bacterial motility. Mol Syst Biol 2007, 3: 128. 10.1038/msb4100166PubMed CentralPubMedView Article
- Braun P, Tasan M, Dreze M, Barrios-Rodiles M, Lemmens I, Yu H, Sahalie JM, Murray RR, Roncari L, de Smet AS, Venkatesan K, Rual JF, Vandenhaute J, Cusick ME, Pawson T, Hill DE, Tavernier J, Wrana JL, Roth FP, Vidal M: An experimentally derived confidence score for binary protein-protein interactions. Nat Methods 2009, 6: 91–97. 10.1038/nmeth.1281PubMed CentralPubMedView Article
- Edwards AM, Kus B, Jansen R, Greenbaum D, Greenblatt J, Gerstein M: Bridging structural biology and genomics: assessing protein interaction data with known complexes. Trends Genet 2002, 18: 529–536. 10.1016/S0168-9525(02)02763-4PubMedView Article
- Moriuchi H, Moriuchi M, Straus SE, Cohen JI: Varicella-zoster virus (VZV) open reading frame 61 protein transactivates VZV gene promoters and enhances the infectivity of VZV DNA. J Virol 1993, 67: 4290–4295.PubMed CentralPubMed
- Cohen JI, Seidel K: Varicella-zoster virus (VZV) open reading frame 10 protein, the homolog of the essential herpes simplex virus protein VP16, is dispensable for VZV replication in vitro. J Virol 1994, 68: 7850–7858.PubMed CentralPubMed
- Che X, Zerboni L, Sommer MH, Arvin AM: Varicella-zoster virus open reading frame 10 is a virulence determinant in skin cells but not in T cells in vivo. J Virol 2006, 80: 3238–3248. 10.1128/JVI.80.7.3238-3248.2006PubMed CentralPubMedView Article
- Che X, Berarducci B, Sommer M, Ruyechan WT, Arvin AM: The ubiquitous cellular transcriptional factor USF targets the varicella-zoster virus open reading frame 10 promoter and determines virulence in human skin xenografts in SCIDhu mice in vivo. J Virol 2007, 81: 3229–3239. 10.1128/JVI.02537-06PubMed CentralPubMedView Article
- Cox E, Reddy S, Iofin I, Cohen JI: Varicella-zoster virus ORF57, unlike its pseudorabies virus UL3.5 homolog, is dispensable for viral replication in cell culture. Virology 1998, 250: 205–209. 10.1006/viro.1998.9349PubMedView Article
- Fuchs W, Klupp BG, Granzow H, Rziha HJ, Mettenleiter TC: Identification and characterization of the pseudorabies virus UL3.5 protein, which is involved in virus egress. J Virol 1996, 70: 3517–3527.PubMed CentralPubMed
- Fuchs W, Granzow H, Klupp BG, Karger A, Michael K, Maresch C, Klopfleisch R, Mettenleiter TC: Relevance of the interaction between alphaherpesvirus UL3.5 and UL48 proteins for virion maturation and neuroinvasion. J Virol 2007, 81: 9307–9318. 10.1128/JVI.00900-07PubMed CentralPubMedView Article
- Rajagopala SV, Titz B, Uetz P: Array-based yeast two-hybrid screening for protein-protein interactions. Yeast Gene Analysis Second edition. 2007, 36: 139–163. [Methods in Microbiology] full_text
- Cagney G, Uetz P: High-throughput two-hybrid screening using yeast arrays. Current Protocols in Protein Science 2001, 19: 1–12.
- Deane CM, Salwinski L, Xenarios I, Eisenberg D: Protein interactions: two methods for assessment of the reliability of high throughput observations. Mol Cell Proteomics 2002, 1: 349–356. 10.1074/mcp.M100037-MCP200PubMedView Article
- Matthews LR, Vaglio P, Reboul J, Ge H, Davis BP, Garrels J, Vincent S, Vidal M: Identification of potential interaction networks using sequence-based searches for conserved protein-protein interactions or "interologs". Genome Res 2001, 11: 2120–2126. 10.1101/gr.205301PubMed CentralPubMedView Article
- Sharan R, Suthram S, Kelley RM, Kuhn T, McCuine S, Uetz P, Sittler T, Karp RM, Ideker T: Conserved patterns of protein interaction in multiple species. Proc Natl Acad Sci USA 2005, 102: 1974–1979. 10.1073/pnas.0409522102PubMed CentralPubMedView Article
- Bartel PL, Chien C-T, Sternglanz R, Fields S: Using the two-hybrid system to detect protein-protein interactions. In Cellular Interactions in Development: A Practical Approach. Edited by: Hartley DA. Oxford: Oxford University Press; 1993:153–179.
- Legrain P, Dokhelar MC, Transy C: Detection of protein-protein interactions using different vectors in the two-hybrid system. Nucleic Acids Res 1994, 22: 3241–3242. 10.1093/nar/22.15.3241PubMed CentralPubMedView Article
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.