Interaction trap cloning with yeast. The following manuscript is a draft of a chapter to appear in "Gene Probes - A practical Approach" published by Oxford University Press. Russell L. Finley Jr. and Roger Brent Department of Molecular Biology Massachusetts General Hospital and Department of Genetics Harvard Medical School Boston Massachusetts 02114 U.S.A. Address: Department of Molecular Biology Massachusetts General Hospital 50 Blossom St. Boston Massachusetts 02114 USA Phone (617) 726-5925 FAX (617) 726-6893 e-mail brent@frodo.mgh.harvard.edu finley@frodo.mgh.harvard,edu Contents 1. Introduction 1.1 Background 1.2 The interaction trap 2. Making and testing baits 2.1 LexA fusion expression plasmids 2.2 Reporters and yeast strains 2.2.1 LEU2 reporter strains 2.2.2 lacZ reporters 2.3 Testing the bait protein 2.3.1 Testing whether the bait protein activates transcription of the reporters Protocol 1. Testing baits for transcription activation 2.3.2 Demonstrating that the bait enters the yeast nucleus and binds operators Protocol 2. The repression assay 2.3.3 Verifying that a full-length fusion protein is made. 3. Libraries 4. An interactor hunt 4.1 Introducing the library into the selection strain. 4.1.1 Selecting interactors from library transformants. 4.1.2 Performing a one step selection for interactors Protocol 3. Transforming the selection strain with library DNA. 4.2 Isolating yeast with galactose dependent Leu+ and lacZ+ phenotypes Protocol 4. Selecting interactors. 5. Verifying specificity Protocol 5. Isolating and classifying library plasmids. Protocol 6. Determining specificity of interactors. 6. Using a mating assay to verify specificity. Figure 4a. Mating assay cartoon. Figure 4b. Mating assay result. Protocol 7. Mating assay. 7. Expected results. Appendix Sequencing and PCR primers for pEG202 and pJG4-5 Media recipes. Interaction trap cloning with yeast. 1. Introduction The interaction trap is a two-hybrid system for cloning cDNAs that encode proteins that interact with a protein whose coding sequences are known. The method uses the transcription of yeast reporter genes as a synthetic phenotype to detect protein-protein interactions. It can also be used to study interactions between known proteins. 1.1 Background The two-hybrid approach takes advantage of the modular domain structure of eukaryotic transcription factors. Many eukaryotic transcription activators have at least two distinct functional domains, one that directs binding to specific DNA sequences and one that activates transcription (1, 2). This modular structure is best illustrated by yeast experiments showing that the DNA-binding domains or activation domains can be exchanged from one transcription factor to the next and retain function. For example, when the DNA-binding domain of the yeast transcription factor Gal4 is replaced with the DNA binding domain of the bacterial repressor LexA, the resulting hybrid protein activates transcription of genes containing upstream LexA binding sites (3). Similarly, when the DNA binding domain of Gal4, which by itself does not activate transcription, is fused to activation domains from other proteins the resulting hybrid proteins activate transcription of reporters with upstream Gal4 binding sites (4-6). A crucial corollary of the modular nature of transcription activators is that the DNA-binding and activation domains need not be covalently attached to each other for activation to occur. This was first demonstrated by Ma and Ptashne (7) with a Gal4 derivative that contained the DNA-binding domain as well as a domain that interacts with another yeast protein, Gal80, but that lacked the activation domain. When this derivative was expressed in yeast it did not activate transcription of a reporter gene containing upstream Gal4 binding sites. However, when it was co- expressed with a second, hybrid protein, consisting of Gal80 fused to an activation domain, interaction between the Gal4 DNA-binding derivative and the Gal80-activation domain hybrid resulted in activation of the reporter gene. The general utility of the modularity of transcription factors was demonstrated by Fields and Song (8) who showed that yeast transcription could be used to assay the interaction between two proteins if one of them was fused to a DNA-binding domain and the other was fused to an activation domain. In their experiment, one of the hybrid proteins contained the DNA-binding domain of Gal4 fused to the yeast protein Snf1, and the other contained the activation domain of Gal4 fused to another yeast protein, Snf4. When Snf1 and Snf4 interacted they brought together the DNA-binding and activation domains, so that the two hybrid proteins bound to Gal4 binding sites upstream of a lacZ reporter gene and activated its transcription. Thus, the interaction between Snf1 and Snf4 was assayed as production of beta-galactosidase. The success of this experiment prompted Fields and Song to make the seminal suggestion that yeast transcription could be used in this way to clone cDNAs encoding proteins that interact with a given known protein (8). In their scheme, a known protein is expressed fused to the DNA- binding domain of Gal4, and a cDNA library is expressed so that proteins encoded by the cDNA are fused to an activation domain (activation-tagged). Transcription of a reporter gene will be activated in yeast containing activation-tagged cDNA-encoded proteins that interact with the known protein. Based on this suggestion, two-hybrid cloning systems have been developed in several labs (9-13). All have three basic components: Yeast vectors for expression of a known protein fused to a DNA-binding domain, yeast vectors that direct expression of cDNA- encoded proteins fused to a transcription activation domain, and yeast reporter genes that contain binding sites for the DNA-binding domain. These components differ in detail from one system to the other. All systems utilize the DNA binding domain from either Gal4 or LexA. The Gal4 domain is efficiently localized to the yeast nucleus where it binds with high affinity to well-defined binding sites which can be placed upstream of reporter genes (14-16). LexA does not have a nuclear localization signal, but enters the yeast nucleus and, when expressed at a sufficient level, efficiently occupies LexA binding sites (operators) placed upstream of a reporter gene (3, 17, 18). No endogenous yeast proteins bind to the LexA operators. Different systems also utilize different reporters. Most systems use a reporter that has a yeast promoter, either from the GAL1 gene or the CYC1 gene, fused to lacZ (19, 20). These lacZ fusions either reside on multicopy yeast plasmids or are integrated into a yeast chromosome. To make the lacZ fusions into appropriate reporters, the GAL1 or CYC1 transcription regulatory regions have been removed and replaced with binding sites that are recognized by the DNA-binding domain being used. A screen for activation of the lacZ reporters is performed by plating yeast on indicator plates that contain X-Gal (5- bromo-4-chloro-3-indolyl-b-D-galactoside); on this medium yeast in which the reporters are transcribed produce beta-galactosidase and turn blue. Some systems use a second reporter gene and a yeast strain that requires expression of this reporter to grow on a particular medium. These "selectable marker" genes usually encode enzymes required for the biosynthesis of an amino acid. Such reporters have the marked advantage of providing a selection for cDNAs that encode interacting proteins, rather than a visual screen for blue yeast. To make appropriate reporters from the marker genes their upstream transcription regulatory elements have been replaced by binding sites for a DNA-binding domain. The HIS3 and LEU2 genes have both been used as reporters in conjunction with appropriate yeast strains that require their expression to grow on media lacking either histidine or leucine, respectively. Finally, different systems use different means to express activation-tagged cDNA proteins. In all current schemes the cDNA- encoded proteins are expressed with an activation domain at the amino terminus. The activation domains used include the strong activation domain from Gal4, the very strong activation domain from the Herpes simplex virus protein VP16, or a weaker activation domain derived from bacteria, called B42. The activation-tagged cDNA-encoded proteins are expressed either from a constitutive promoter, or from a conditional promoter such as that of the GAL1 gene. Use of a conditional promoter makes it possible to quickly demonstrate that activation of the reporter gene is dependent on expression of the activation-tagged cDNA proteins. Many of these systems now provide the investigator with a relatively good chance to recover proteins that interact with other proteins. Because most are based on the same concepts, some of their components are often interchangeable. However, different systems utilize the yeast selectable markers in different ways. Moreover, systems that employ the DNA-binding domain of Gal4 must use a yeast strain that lacks wild type Gal4; these system cannot use library vectors that direct synthesis of the activation-tagged proteins from the GAL1 promoter whose transcription requires Gal4. 1.2 The interaction trap The interaction trap is an implementation of the two-hybrid system developed by Gyuris et. al (11). It consists of three critical components (see Figure 1). First, it uses a vector for expression of a protein of interest fused to LexA. Because the goal of interaction trap cloning is to find proteins that interact with the protein fused to LexA, this hybrid is referred to as the "bait". Second, the trap uses a yeast strain with two reporter genes. One reporter is a yeast LEU2 derivative that has its normal upstream regulatory sequences replaced with LexA operators. Transcription of the LexA-operator- LEU2 gene (LexAop-LEU2) can be measured by the ability of the strain to grow in the absence of leucine, which requires the LEU2 gene product. The LexAop-LEU2 gene is integrated into the yeast chromosome. The other reporter gene is lacZ, which provides a secondary assay of activation by the bait and activation-tagged proteins interacting with it, as well as some quantitative information about the interaction. Third, the interaction trap uses a library plasmid that directs the conditional expression of cDNA-encoded proteins fused at their amino termini to a moiety containing three domains: a nuclear localization signal, a transcription activation domain, and an epitope tag. The activation-tagged cDNA-encoded protein is expressed from the yeast GAL1 promoter, which is induced by galactose and repressed by glucose. The interaction trap is illustrated in Figure 1. The bait protein is constitutively expressed. It binds to LexA operators upstream of the reporter genes LEU2 and lacZ but does not activate their transcription. The activation-tagged cDNA-encoded protein is conditionally expressed from the GAL1 promoter. In glucose medium the GAL1 promoter is repressed, no cDNA-encoded protein is made, and the yeast does not grow in the absence of leucine. When the yeast are grown on galactose medium, activation-tagged cDNA-encoded proteins are expressed, and those that interact with the bait activate transcription of the LEU2 and lacZ reporters. Thus, cells containing activation-tagged cDNA proteins that interact with the bait form colonies on galactose medium lacking leucine and form blue colonies on galactose X-Gal plates. (Figure 1. The interaction trap) An outline of an interactor hunt is presented in Figure 2. The protocols for using the interaction trap described below require knowledge of a few basic yeast microbiological and genetic techniques. A more detailed description of such techniques, together with recipes for appropriate media can be found elsewhere (21-23). (Figure 2. Flow chart of an interactor hunt.) 2. Making and testing baits 2.1 LexA fusion expression plasmids To make a plasmid that directs the synthesis of the LexA fusion or "bait" protein, the coding region for the protein of interest is inserted into pEG202 or a related plasmid (11) (see Appendix). pEG202 is a multicopy yeast plasmid containing the yeast 2 mm origin of replication and the selectable marker gene HIS3, as well as the full-length LexA coding region flanked by the yeast ADH1 promoter and terminator. Bait proteins expressed from this plasmid contain amino acids 1 to 202 of LexA, which include the DNA-binding and dimerization domains. Downstream of the LexA coding region in pEG202 are unique EcoRI, BamHI, SalI, NcoI, NotI, and XhoI cloning sites. The bait plasmid can be introduced and maintained in a his3 yeast strain (e.g. EGY48, see below) by selecting transformants on media lacking histidine. Transformants will constitutively express the protein of interest with LexA at its amino terminus. Although it does not contain a yeast nuclear localization signal, LexA and most LexA fusions will enter the nucleus (3, 17, 18, 24-29). The expression levels afforded by the ADH1 promoter are generally sufficient to provide occupancy of LexA operators upstream of the reporter genes. For the rare bait that is excluded from the nucleus, a pEG202 derivative can be used that directs expression of LexA fusions that contain a nuclear localization signal (W. Breitwieser and A. Ephrussi, personal communication). 2.2 Reporters and yeast strains 2.2.1 LEU2 reporter strains An interactor hunt employs a selection for cDNAs encoding interactors. The selection depends on a yeast strain, EGY48 (11), that has an integrated LEU2 gene with its upstream regulatory region replaced by LexA operators. The strain has no other LEU2 gene and does not grow in the absence of leucine unless the LexAop-LEU2 gene is transcribed. The LEU2 reporter in EGY48 is very sensitive; it is activated by even weak transcription activators fused to LexA, or by activation-tagged proteins that interact weakly with LexA fusions. The high sensitivity is due to the presence of three high affinity LexA operators positioned near the LEU2 transcription start. The operators are from the bacterial colE1 gene and each can potentially bind two LexA dimers (30). While the sensitivity of EGY48 is an advantage in that it facilitates isolation of activation-tagged cDNA proteins that interact weakly with the bait, this strain may be too sensitive for use with baits that are themselves weak transcription activators. Many proteins, including some that are not transcription factors, will activate transcription of LEU2 in EGY48. For a bait to be used in an interactor hunt it must not activate LEU2 transcription. For baits that fail to meet this criterion, two approaches can be taken. First, the sensitivity of the reporter strain can be reduced. One way to do this is by using a strain containing fewer operators upstream of LEU2 (e.g., one operator instead of three; E. Golemis, D. Krainc, R.L.F., unpublished data). If a bait still activates transcription of LEU2 in a strain with only one operator, the sensitivity can be reduced further by using a diploid yeast strain, in which, for unknown reasons, LexA fusions activate less transcription of the reporter genes (E. Golemis, A. Mendelsohn, D. Krainc, R.L.F., unpublished data). A second, more drastic, approach is to construct deletion derivatives of it that do not activate. A good way to start, if prior knowledge of the precise location of transcription activation domains is unavailable, is to construct derivatives that lack highly acidic regions which are often responsible for transcription activation in yeast (2, 4, 31). The obvious disadvantage of this approach is that regions important for interaction with other proteins may be removed. In addition to the mutation in the endogenous LEU2 gene, EGY48 and related strains carry mutations in three other marker genes (his3, trp1, ura3) that are needed to allow selection of the plasmids used in the interaction trap. The his3 mutation is complemented by the HIS3 gene on the bait expression vector. The trp1 and ura3 mutations are respectively complemented by the TRP1 gene on the library plasmid, and the URA3 gene on the lacZ reporter plasmid (see below). The plasmids for the bait, library, and lacZ reporter each contain the yeast 2 mm origin of replication so that under continued selection they should be maintained at 20 to 100 copies per cell (32). 2.2.2 lacZ reporters In addition to the LexAop-LEU2 reporter, an interactor hunt employs a LexAop-lacZ reporter. The lacZ reporters contain the GAL1 TATA, transcription start, and a small part of the GAL1 coding sequence fused to lacZ (19, 33). The GAL1 upstream activating sequences (UASg) have been replaced with an XhoI site into which various numbers of LexA operators have been inserted (see Appendix). In the absence of a LexA fusion, or interacting activation-tagged protein, yeast bearing these reporters make no detectable beta-galactosidase and appear white on X-Gal plates. Use of the lacZ reporters provides two advantages in an interactor hunt. First, false positives that may arise by activation of the LEU2 gene due to a yeast mutation, or to binding of the activation-tagged cDNA protein to the LEU2 promoter, can be identified because they will fail to activate the lacZ reporter. Second, the lacZ reporters provide a relative measure of the amount of transcription caused by the interaction of an activation-tagged cDNA protein with a bait. The phenotype measured with the LEU2 reporter, growth in the absence of leucine, while very sensitive, is difficult to quantitate. In contrast, the beta-galactosidase activity in a yeast is directly proportional to the amount of lacZ transcription, and is easily measured (33). Careful use of the lacZ reporters may even allow comparison of interaction affinities between different baits and activation-tagged proteins (11). The threshold affinity of protein-protein interactions to be detected in an interactor hunt can be adjusted by choosing between different lacZ reporters. The sensitivity of the lacZ reporter phenotype depends on the number of LexA operators positioned upstream of lacZ. Activation-tagged proteins that interact weakly with a bait can be identified by using a more sensitive lacZ reporter containing more operators. However, the search can be limited to find only cDNA-encoded proteins that interact tightly with the bait by using a less sensitive lacZ reporter. All lacZ reporters commonly used are less sensitive than LexAop-LEU2 reporters. Because of this, some LexA activators will activate LEU2 and allow EGY48 to grow in the absence of leucine, but will not activate lacZ and cause the strain to turn blue on X-Gal plates. The lacZ reporters reside on 2 mm plasmids containing the URA3 gene (see appendix). 2.3 Testing the bait protein Before conducting an interactor hunt, the bait should be tested to ensure that the fusion protein enters the nucleus, binds LexA operators, and does not activate transcription of the reporter genes. This is done in two steps. First, a selection strain is made by introducing the bait plasmid into EGY48 that contains a LexAop-lacZ reporter. The resulting selection strain is used to show that the bait protein does not activate transcription of LEU2 and lacZ (Protocol 1). Eventually, the library will be introduced into this strain. Second, the bait plasmid is introduced into a related strain that contains a different lacZ reporter to verify that the bait protein enters the yeast nucleus and binds LexA operators. This is done with a repression assay (Protocol 2). 2.3.1 Testing whether the bait protein activates transcription of the reporters Protocol 1 describes how to verify that the bait does not activate transcription of the LEU2 reporter. In addition to the bait expression plasmid, this protocol uses three related HIS3 plasmids as controls. The first is a plasmid that makes no LexA protein, or one that makes LexA fused to a protein that does not activate transcription. EGY48 derivatives that contain such plasmids fail to grow on media lacking leucine. The second is a plasmid that makes LexA fused to a transcription activation domain, like the activation domain of Gal4. Such a plasmid will allow EGY48 to grow in the absence of leucine. The third is the parent plasmid, pEG202. The LexA protein encoded by pEG202 includes several amino acids encoded by the polylinker which make the protein a weak transcription activator. EGY48 containing pEG202 grows slowly on medium lacking leucine and eventually forms colonies. A good criterion for determining whether a bait plasmid can be used for an interactor hunt is to show that it causes EGY48 to grow more slowly on medium lacking leucine than pEG202. Note on yeast transformation. While only a handful of yeast transformants are needed in Protocol 1 and Protocol 2, the transformation efficiency (transformants per mg of plasmid) must be very high for Protocol 3, in which the selection strain is transformed with the library. To become familiar with high efficiency yeast transformation is advisable to use a high efficiency method for the transformations in Protocols 1 and 2. There are several effective high efficiency yeast transformation protocols to choose from including electroporation (34, 35) and those that employ lithium salts (36). The method described in Protocol 3 results in about 10e5 transformants per mg and may be scaled down to use in Protocol 1 and 2. Two other considerations deserve mention. First, once a plasmid has been introduced into a strain it must be maintained by continued selection for its presence. Thus, a strain that already contains the URA3 lacZ reporter plasmid should be transformed with a second plasmid by first growing on media lacking uracil (Glu ura-; see appendix for media designations). Although strains that contain different plasmids can be constructed by introducing more than one plasmid at a time, the transformation efficiency will be lower than when strains are constructed by serial transformation of one plasmid at a time. Second, each time a strain is transformed, a control transformation should be performed using no plasmid DNA. ______________________________________________________ Protocol 1. Testing baits for transcription activation 1. To construct the selection strain, transform EGY48 with a URA3 lacZ reporter plasmid and select transformants on Glu ura- plates (see appendix for plate designations; see above note on transformation). Combine three colonies from these plates, grow in Glu ura- liquid, and transform them with the HIS3 bait plasmid (or control plasmids). Select transformants on Glu ura-his- plates. 2. Pick four individual colonies from each transformation and use each colony to inoculate 5 ml Glu ura-his- liquid cultures. At the same time, streak the same four transformants to another Glu ura- his- plate for storage and later recovery. All four should behave identically in the tests below, in which case any one will serve as the selection strain into which the library will be introduced. Incubate these plates at 30oC until colonies form (about two days) and then store at 4oC. 3. Grow the liquid cultures at 30oC shaking to OD600=0.5 (corresponding to about 10e7 cells/ml). This is mid-log phase. If the overnight cultures grow to a density greater than OD600=0.5, dilute to less than OD600=0.2 and then grow to OD600=0.5 so that the cells are in mid-log phase when harvested. 4. Make 10e2- and 10e3 -fold dilutions of each culture in sterile water. Spot 10 ml of the culture and 10 ml of the dilutions onto two plates: ** Gal/Raf ura-his- ** Gal/Raf ura-his-leu- Spot yeast containing the bait plasmid being tested and yeast containing the control plasmids onto the same plates for side by side comparison. Incubate at 30oC. Note: galactose is used in the media because the actual selection will eventually be done on galactose plates to induce expression of the activation-tagged cDNA protein. Raffinose is added to aid yeast growth; it provides a better carbon source than galactose alone but does not block the ability of galactose to induce the GAL1 promoter (R.L.F., unpublished data). 5. Monitor the growth of the cells in the spots for several days. The yeast in all of the spots should grow at a similar rate on the Gal/Raf ura-his- plates. If growth on the Gal/Raf ura-his- plates is reproducibly diminished for a given bait, relative to the controls, it may indicate that its expression is toxic to yeast. Yeast with no LexA or with a non-activating LexA fusion should not grow after several days on Gal/Raf ura-his-leu- plates. Yeast containing LexA fused to a protein that activates transcription may grow as fast on Gal/Raf ura- his-leu- plates as on Gal/Raf ura-his- plates, depending on the strength of the activator. The above steps establish whether a bait can be used in an interactor hunt. For a bait to be used, the selection strain containing it must not grow on Gal/Raf ura-his-leu- plates for two to three days. If yeast that contain a bait begin form visible colonies before three days, it may be necessary to construct a less sensitive selection strain (see section 2.2.1). However, if the yeast grow very slowly, e.g. form colonies only after three days, it is still possible to do an interactor hunt by selecting yeast with library plasmids that cause colonies to form in one or two days. 6. Look for lacZ expression in the selection strain. Patch individual transformants from step 1 to Glu ura-his- X-Gal plates and incubate at 30oC. Yeast with the control LexA-activator fusion should turn blue overnight while those lacking LexA or containing a transcriptionally inert bait will remain white after many days. If a bait activates the LEU2 gene but not the lacZ gene (which is frequently observed because the LexAop-LEU2 reporter is more sensitive than LexAop-lacZ reporter), it may be possible to perform an interactor hunt by screening for yeast containing activation- tagged cDNA proteins that interact with the bait and activate the lacZ reporter; these yeast will be blue on X-Gal plates. However, this method loses the advantage of a selection. ______________________________________________________ 2.3.2 Demonstrating that the bait enters the yeast nucleus and binds operators Baits that do not activate the LEU2 reporter in the assay in Protocol 1 should be tested to be sure they enter the nucleus and bind to LexA operators. This can be done with a repression or "blocking" assay. The repression assay is based on the observation that LexA and non-activating LexA fusions can repress transcription of a yeast reporter gene that has operators positioned between the TATA and upstream activating sequence (UAS) (18). The mechanism of this repression is not understood but presumably is not equivalent to repression by repressor proteins native to yeast (37). While some LexA fusions repress more than others (24), repression does depend on the presence of operators in the reporter, and thus any repression observed may be attributed to operator occupancy by the bait. The reporter plasmid used for the repression assay (pJK101; see Appendix) is similar to the plasmids used to test activation; it contains the 2 mm origin, URA3, and a GAL1-lacZ fusion. Unlike the plasmids used for testing activation, the GAL1-lacZ fusion in pJK101 contains most of the GAL1 upstream activating sequence, UASg. In addition it contains one LexA operator positioned between UASg and the TATA box. lacZ expression is induced by galactose and is detectable in the presence of glucose, because negative regulatory elements that normally keep GAL1 completely repressed in glucose are not present (38). Transcriptionally inert LexA fusions that bind to the operator in pJK101 repress lacZ expression from 2 to 20-fold in the presence of galactose. Repression appears more profound when the yeast are grown in glucose medium because there is less lacZ expression to begin with. The repression assay can often be done on X-Gal plates by looking for differences in blueness between yeast with different baits. However, when looking for low levels of lacZ expression in yeast grown in glucose, or when looking for slight differences in lacZ expression (2 to 4-fold), more sensitive beta- galactosidase assays may be necessary (39, 40). ______________________________________________________ Protocol 2. The repression assay 1. Transform EGY48 with pJK101 and select transformants on Glu ura- plates. Combine three colonies from these plates and transform them with the HIS3 bait plasmid (or HIS3 control plasmids). Select transformants on Glu ura-his- plates. 2. Pick four individual colonies from each transformation and streak a patch of them onto Glu ura-his- and Gal/Raf ura-his- plates containing X-Gal. Incubate at 30oC. 3. Examine the X-Gal plates after 1, 2, and 3 days. Yeast lacking LexA will begin to turn blue on the Gal/Raf plates after one day and will appear light blue on the glucose plates after two or more days. Yeast containing a bait that enters the nucleus and binds operators turn blue more slowly than the yeast lacking LexA. 4. Baits that repress transcription of lacZ in pJK101 by 2-fold or less may not cause a visible reduction in blue on X-Gal plates. If no repression is observed on the X-Gal plates, perform beta-galactosidase assays with transformants from step 1. Grow the transformants in 5 ml Glu ura-his- and Gal/Raf ura-his- liquid media, or on Glu ura-his- and Gal/Raf ura-his- plates for 2 days, before doing beta-galactosidase assays (39, 40). ______________________________________________________ 2.3.3 Verifying that a full-length fusion protein is made. Finally, it is usually good practice to demonstrate that the full- length bait protein is made. This can be done by running extracts from yeast cells that harbour the bait plasmid on an SDS polyacrylamide gel, immunoblotting with either an antibody to LexA or one specific to the protein fused to LexA (27, 29), and detecting a fusion protein of the expected apparent molecular weight. Yeast cell extracts can be prepared by growing yeast in liquid culture (lacking histidine to maintain selection for the bait plasmid) to OD600 of 0.5, spinning 1 ml of the culture to pellet the cells, and resuspending the cells in 50 ml of 2X Laemmli sample buffer (41). The cells can then be broken by freezing on dry ice followed by boiling for 5 minutes prior to loading on an SDS polyacrylamide gel (about 15 ml/lane). The proteins can then be transferred to a filter and blotted with standard immunoblotting (western) methods (22, 42). 3. Libraries In an interactor hunt the expressed cDNA-encoded proteins are fused to an activation domain, as well as a nuclear localization signal to increase their nuclear concentration, and an epitope tag so they may be immunologically identified. The prototypical library plasmid for expression of these activation-tagged cDNA proteins is pJG4-5 (11) (see Appendix). This is a 2 mm plasmid that contains the TRP1 marker and the GAL1 promoter. Downstream of the GAL1 promoter there is an ATG followed by 105 codons. These encode 9 amino acids from the SV40 Large T nuclear localization signal, 87 amino acids that make up the activation domain called B42, followed by 9 amino acids comprising the hemagluttinin (HA) epitope tag. The B42 domain is derived from E.coli and acts as a moderately strong transcription activation domain in yeast (4). Use of this activation domain avoids the possible toxic effects of overexpressing a strong activation domain (43, 44). Downstream of the sequences that encode the fusion moiety there are unique EcoRI and XhoI sites for insertion of cDNAs. Proteins encoded by cDNAs that are in-frame carry the fusion moiety at their amino terminus. The activation- tagged cDNA-encoded proteins will be expressed in yeast grown on galactose but not in yeast grown in glucose. Numerous libraries have been made using pJG4-5. These include cDNA libraries made from RNA derived from HELA cells (11), Drosophila, ovaries, discs, and 0 to 12 h embryos (Finley and Brent, in preparation), adult Drosophila heads (J.Huang and M. Rosbach, personal communication), Drosophila 16 to 26 h embryos (V. Neel and M. Young, personal communication), serum-starved WI38 cells (C. Sardet, J. Gyuris, R.B., unpublished data), human brain (D. Krainc and R.B., unpublished data), Arabidopsis (H.Zhang and H.Goodman, personal communication), and a library made from yeast genomic DNA (P. Watt personal communication). Construction of libraries is a complex topic beyond the scope of this article, but is described elsewhere (11, 45, 46) (Finley and Brent, in preparation). 4. An interactor hunt 4.1 Introducing the library into the selection strain. 4.1.1 Selecting interactors from library transformants. To conduct an interactor hunt the library is introduced into the selection strain in a large transformation so that many transformed cells are obtained, each of which contains an individual library plasmid. Those cells that contain library plasmids that encode proteins that interact with the bait protein are then selected. To date, the most frequently successful method has been to do this in two steps. In the first step, yeast that already contain the bait and lacZ plasmids are transformed with the library and library transformants are isolated and frozen. In the second step, the selection for interactors is applied to the library transformants. In this approach yeast transformed with the TRP1 library plasmid are selected by plating the transformation mix onto medium lacking tryptophan (that also lacks uracil and histidine to maintain selection for the bait and lacZ plasmids, i.e., Glu ura-his-trp- plates). The transformants are then scraped from the plates and frozen for storage. Aliquots are thawed and plated onto a medium containing galactose and lacking leucine. This induces expression of the activation-tagged cDNA proteins, and selects for transcription of the LEU2 reporter. This two step approach results in a uniform increase in the number of cells carrying each library plasmid; each cell transformed with the library is allowed to multiply to form a colony, and the colonies are harvested at approximately the same size. One advantage to this approach is that the number of transformed yeast is amplified before the synthesis of cDNA-encoded proteins is induced. This ensures that yeast containing toxic or mildly toxic cDNA-encoded proteins will not be depleted from the population. This method is described in Protocol 3 and Protocol 4. 4.1.2 Performing one step selection for interactors A simpler, but so far inferior, alternative approach is to perform a one step selection for yeast containing activation-tagged cDNA proteins that interact with the bait. This method involves plating the library transformation mix directly on plates lacking leucine (i.e. Gal/Raf ura-his-trp-leu- plates) to select for expression of the LEU2 reporter, without first selecting cells transformed with the library plasmid. Although it is easier that the two step method, it suffers from some disadvantages. First, only a fraction of the yeast containing activation-tagged proteins that interact with the bait survive on the leu- selection plates, possibly because some cells die in the time it takes to induce synthesis of the activation-tagged protein, the LEU2 product, and leucine. This lower plating efficiency on the leu- selection plates is particularly evident when the yeast contain an activation-tagged cDNA-encoded protein that interacts only weakly with the bait. Second, the one step approach may reduce the probability of isolating cDNAs that encode proteins that are somewhat toxic to yeast because these proteins will further reduce the plating efficiency on leu- plates. Although both of these disadvantages might be overcome by plating very large numbers of library transformants, obtaining such numbers would require exceptional yeast transformation efficiencies or huge amounts of library DNA. Furthermore, if expression of an activation-tagged cDNA protein causes yeast to plate at reduced efficiency, a much larger number of yeast will need to be characterized to find them. For these reasons we prefer the two step method. However, we also present the best current version of the one step method, which may eventually be improved to overcome the relative disadvantages described above. To do the one step method follow Protocol 3 steps 1 to 10 and then skip to Protocol 4. ______________________________________________________ Protocol 3. Transforming the selection strain with library DNA. The following protocol is a variation of the high efficiency lithium acetate method developed by Geitz et al. (36). Before transforming the strain with library DNA (which is usually fairly valuable), perform pilot transformations of the strain with the TRP1 library vector. A transformation efficiency of at least 10e5 transformants/mg DNA should be the goal. Other high efficiency yeast transformation protocols, e.g. electroporation (34, 35), may be substituted. 1. Grow yeast containing the bait and lacZ reporter plasmids in 400 ml of Glu ura-his- medium at 30oC, with shaking (~150 rpm) to an OD600 of 1.0, corresponding to about 3 X 10e7 cells/ml. The doubling time in this medium is rather slow (2 to 4 hours). For this reason it is sometimes convenient to start a smaller, 50 ml culture, grow it to OD600=2.0 or greater, and then dilute it into 400 ml. For high transformation efficiencies it is important to start the 400 ml culture at OD600=0.2 or less and allow it to grow to OD600=1.0 so that the cells are in mid-log phase when harvested. 2. Spin culture at 2000 X g for 5 min and pour off supernate. Wash the cells in 20 ml sterile water. These spins and all following manipulations are done at 20 - 25oC. 3. Resuspend yeast in 5 ml of filter sterilized LiOAc/TE (10 mM Tris HCl pH7.5, 1 mM EDTA, 100 mM LiOAc; make from a filter sterile stock of 1 M LiOAc, pH 7.5). Pellet again and pour off supernate. 4. Resuspend yeast in 2.0 ml LiOAc/TE. 50 ml of this suspension provides enough competent cells to be transformed with 1 mg of DNA. Note: use of more than 1 mg DNA per 50 ml of cells can result in introduction of multiple library plasmids in each yeast cell. This should be avoided because of the large amount of additional work that will be required to determine which of the library plasmids in a cell expresses the cDNA-encoded protein that interacts with the bait (see below). 5. Aliquot 100 ml of competent yeast into sterile eppendorf tubes. To each tube add 2 mg of library DNA and 60 mg of carrier (single stranded salmon sperm; (47)) in a total volume of 20 ml. The number of tubes to be used depends on the desired number of transformants and the expected transformation efficiency. If pilot transformations resulted in efficiencies of 10e5 transformants per mg, each tube should yield 200,000 transformants, all of which can be plated on onto a single 24cm X 24cm plate (see below). 6. Add DMSO to 10% vol/vol. This improves transformation efficiency by 3 to 5-fold (48) (R.L.F. and B. Cohen, unpublished data). 7. Add 600 ml of filter sterilized 40% PEG 4000 in LiOAc/TE (made from stocks of 1 M LiOAc pH 7.5, filter sterile 50% PEG 4000 in water, 1 M Tris HCl pH 7.5, and 0.5 M EDTA). Gently invert tube several times to mix. 8. Incubate at 30oC for 30 min. Agitation is not necessary. 9. Heat shock at 42oC for 15 min and return to room temperature. 10. Determine the total number of transformants by removing 10 ml from each tube and making three dilutions (10-, 10e2-, and 10e3- fold) in sterile water. Plate 100 ml of each dilution onto 100 mm Glu ura-his-trp- plates and incubate at 30oC. You will be able to calculate the total number of transformants from the number of colonies on these dilution plates. At this point the transformation mixes can be plated to first select for all library transformants (as discussed in 4.1.1). This method is described below starting with step 11. Alternatively, the transformation mix can be used in a one step selection for interactors (as discussed in 4.1.2); to perform this, proceed to Protocol 4. Selecting library transformants. 11. Plate each transformation mix (less then 1 ml) onto a single 24cm X 24cm Glu ura-his-trp- plate. There is no need to spin the cells or remove the PEG. The media in these plates should be at least 0.6 cm thick, level, and free of bubbles. To achieve an even distribution of cells pour about 100 sterile glass beads (4 mm diameter, Fisher Scientific; sterilized by autoclaving) on the plate with the cells. Gently roll the beads around the plate to distribute the transformation mix, then pour the beads off, or onto the next plate. Alternatively, distribute the transformation mix with a sterile bent glass rod. Both techniques work best when the surface of the plates are not too wet so that the media absorbs the transformation mix. To achieve this moisture content put newly solidified plates into a laminar flow hood with the lids ajar for about two hours before plating. 12. Incubate the plates at 30oC. Colonies should appear after about 24 hours. Continue incubation until colonies are 1 to 2 mm in diameter, which should take a total of approximately 2 days. Harvest the transformants. 13. Place the plates at 4oC for 2 to 4 hours to harden the agar. 14. Using the long edge of a sterile glass microscope slide (and sterile technique) scrape the yeast from the plate. Try not to scrape any agar as it will interfere with pipeting. Collect the yeast from the glass slide by wiping it on the lip of a sterile 50 ml Falcon tube. 16. Wash the cells two times with 2 or 3 volumes of sterile TE. It is best to pellet the cells each time in a sterile round bottom polypropylene tube at 2000 X g for 4 min so they may be easily resuspended. The pellet volume for 500,000 transformants will be about 8 ml. 17. Resuspend the cells thoroughly by swirling in 1 pellet volume of glycerol solution (65% glycerol (vol/vol), 0.1 M MgSO4, 25 mM Tris pH 7.4). Mix well by vortexing on low speed. Freeze 1 ml aliquots at -70oC. 18. Determine the plating efficiency by thawing an aliquot of library transformants and making serial dilutions in sterile water. Plate 100 ml of each dilution onto 100 mm Gal/Raf ura-his-trp- plates. Count the colonies that grow after 2 to 3 days at 30oC and represent the plating efficiency in colony forming units (CFU) per unit volume of frozen cells. The plating efficiency will be on the order of 10e8 CFU/100 ml. ______________________________________________________ 4.2 Isolating yeast with galactose dependent Leu+ and lacZ+ phenotypes Library transformants containing cDNAs that encode proteins that interact with the bait will exhibit galactose-dependent growth on media lacking leucine (Leu+), and galactose-dependent beta- galactosidase activity (lacZ+). Isolation of these galactose-dependent Leu+ and lacZ+ yeast is accomplished in two steps. First, library transformants are plated on galactose medium lacking leucine (leu-) to select yeast that are Leu+. Second, the Leu+ yeast are isolated, placed on a glucose master plate, and then replica plated to four new plates to test for lacZ expression and galactose dependence. These four plates include two leu- plates and two X-Gal plates: one leu- plate and one X-Gal plate contain galactose to induce cDNA expression (plus raffinose to enhance growth), while the other leu- plate and the other X-Gal plate contain glucose to repress cDNA expression. Yeast that grow on leu- galactose medium but not on leu- glucose medium, and that turn blue on galactose X-Gal plates but remain white on glucose X-Gal plates (i.e. those that are galactose- dependent Leu+ and lacZ+) are picked for further characterization. In this procedure the yeast are grown on the glucose master plates to shut off cDNA expression before replica plating so that the galactose dependence of the Leu+ and lacZ+ phenotypes can be assessed. This avoids a problem that can arise when the Leu+ yeast are replica plated from galactose to glucose plates: There may be sufficient message and protein product from the activation-tagged cDNA protein to allow the yeast to grow on leu- glucose for several generations and turn blue on glucose X-Gal without further cDNA expression. This would mask the galactose-dependence of the Leu+ and lacZ+ phenotypes. ______________________________________________________ Protocol 4. Selecting interactors. Note: If TRP+ library transformants were selected and stored frozen follow "a" in the steps below. If library transformants were not selected and stored, follow "b" in the steps below. 1. Induce synthesis of activation-tagged cDNA-encoded proteins. a) If library transformants were selected and frozen, thaw an aliquot of the cells and dilute 10-fold into Gal ura-his-trp- liquid media. b) If performing the one step method to select for interactors, dilute the transformation mix from Protocol 3 step 9, 10-fold into Gal ura-his-trp- media. Incubate at 30oC with shaking for 4 hours to induce the GAL1 promoter. There is almost no increase in cell number during this time, and any increase can be neglected when calculating the number of colony forming units or transformants to plate onto leu- selection plates. 3. Pellet cells at 2000 X g for 4 min at 20 - 25oC and resuspend in sterile water. 4. Plate onto Gal/Raf ura-his-trp-leu- plates. a) If library transformants were pre-selected and frozen, plate 10e6 CFU (determined from the plating efficiency test in Protocol 3 step 17) onto each 100 mm plate. b)If performing the one step selection, plate an amount of the transformation mix that should contain 10e6 transformants, based on estimates from transformation efficiencies obtained in pilot transformation experiments, onto each 100 mm plate. Determine the actual number of library transformants plated by counting the colonies that grow when dilutions of the transformation mix are plated onto Gal/Raf ura-his-trp- as described in Protocol 3 step 10. 5. Incubate selection plates at 30oC. Colonies should appear in 2 to 5 days. To keep the plates from drying out after two days, it may be helpful to put them in plastic bags or containers, or put parafilm around each plate. Generally, there will be more galactose- dependent Leu+ and lacZ+ yeast among the colonies that appear sooner, and fewer among the colonies that appear later. 6. Pick colonies with sterile toothpicks or applicator sticks and patch, or streak for single colonies, onto another Gal/Raf ura-his-trp-leu- plate. Ideally the Leu+ yeast should be streaked for single colonies to isolate them from contaminating Leu- yeast that were present when the Leu+ colony was forming. However, when there are large numbers of Leu+ colonies, it may be inconvenient to streak purify every one; in this case growth of patches on a second selection plate will at least enrich for the Leu+ cells. 7. To show that the Leu+ phenotype is galactose-dependent, patch the Leu+ yeast onto Glu ura-his-trp- master plates to turn off the GAL1 promoter and stop expression of the activation-tagged cDNA protein. Grow at 30oC for about 24 hours. 8. Replica the master plates to the following five plates: 1. Glu ura- his-trp- X-Gal; 2. Gal/Raf ura-his-trp- X-Gal; 3. Glu ura-his-trp-leu-; 4. Gal/Raf ura-his-trp-leu-. Incubate at 30oC and examine results after 1, 2, and 3 days. Pick yeast that are Leu+ and lacZ+ only on galactose (for example, see Figure 3). Further characterize these by isolating the library plasmid and determining the interaction specificity. ______________________________________________________ Three alternative procedures to the method described in Protocol 4 have been used successfully. First, if a large number of Leu+ yeast colonies appear on the initial leu- selection plates, those that are lacZ+ can be quickly identified using a filter beta-galactosidase assay. The filter assay, described in detail elsewhere (49), involves lifting a replica of the yeast from the colonies on the leu- selection plate with a nitrocellulose filter, lysing the yeast on the filter, and exposing the filter to buffer and X-Gal. The filter is then examined for blue spots corresponding to lacZ+ yeast. Leu+ yeast that correspond to those that are also lacZ+ are then picked from the original leu- selection plate, put onto glucose master plates, and replica plated as described in Protocol 4, steps 7 and 8. The second alternative is to include X-Gal in the initial leu- selection plates and pick those that grow and turn blue. Again, the Leu+ lacZ+ yeast are picked, patched onto glucose master plates, and replica plated to the two sets of indicator plates. The disadvantage to this approach is that yeast grow less well on X-Gal plates because these plates contain a buffer to give them neutral pH; the reduced growth rate at this higher pH reduces the plating efficiency during the selection. Finally, the yeast colonies on the original leu- selection plate can be replica plated directly to X-Gal plates to determine which are lacZ+ (A. Mendelsohn, personal communication). Only those that are Leu+ and lacZ+ are isolated and tested for galactose dependence. Most activation-tagged proteins that interact strongly with the bait will render the yeast containing them galactose-dependent Leu+ and lacZ+. However, yeast containing activation-tagged cDNA proteins that interact only weakly with the bait may be galactose- dependent Leu+, yet may appear light blue or even white on X-Gal plates. This is due to the different sensitivities of the Leu and lacZ phenotypes (see 2.2). Results from several interactor hunts have shown that there are usually more biologically relevant interactors in the class of yeast that are galactose-dependent Leu+ and lacZ+ than in yeast that are lacZ- (R.L.F., unpublished data). However, the class of yeast that are galactose-dependent Leu+ but lacZ- may also contain biologically relevant interactors (Finley and Brent, in preparation). For this reason the decision whether or not to further characterize the latter class must be somewhat arbitrary. 5. Verifying specificity A finding of galactose-dependent Leu+ and lacZ+ in a yeast isolate can be considered a demonstration that the reporter genes are activated due to expression of the activation-tagged cDNA-encoded protein. However, it is important to determine that activation of the reporters is due to specific interaction of this protein with the bait, rather than to its nonspecific interaction with LexA, with the promoters, or with some part of the transcription machinery. To verify that the cDNA-encoded protein interacts specifically with the bait, library plasmids are rescued from the galactose-dependent Leu+ lacZ+ yeast and re-introduced into the original selection strain and into other strains containing different baits. Specific interactors confer the galactose-dependent Leu+ and lacZ+ phenotype to yeast containing the original bait, but not to yeast containing unrelated baits. To test this, master plates are made from the new transformants and replica plated onto four indicator plates as described in Protocol 4 steps 7 and 8 (see Figure 3). Alternatively, the specificity can be determined using a mating assay (Protocol 6). Library plasmids are rescued from yeast by performing a quick yeast plasmid miniprep and using the miniprep DNA to transform E.coli (most yeast miniprep protocols do not provide enough clean plasmid DNA for restriction analysis). If a large number of galactose- dependent Leu+ lacZ+ yeast are obtained, it is useful to reduce the number of library plasmids that need to be rescued by determining which ones contain identical cDNAs. This can be done by comparing restriction digests of PCR products containing the cDNA insert. In this procedure, yeast miniprep DNA is used as template in PCR reactions with primers derived from sequences in the library plasmid flanking the cDNA insertion site. PCR products are then digested with one or two restriction enzymes that cut frequently. This procedure is described in Protocol 5. Since the yeast contain three different plasmids, rescuing the library plasmid depends on distinguishing it from the bait and lacZ reporter plasmids. There are at least three ways to do this. In the most efficient one, yeast minipreps are used to transform a strain of E.coli that contains a mutation in the trpC gene. The inability of the trpC- E.coli to grow in the absence of tryptophan is complemented by the yeast TRP1 gene on the library plasmid (50). trpC E.coli transformed with the TRP1 library plasmid are selected on minimal plates lacking tryptophan and containing ampicillin. In another method, yeast minipreps are used to transform E.coli, and E.coli minipreps from several individual transformants are analyzed by restriction digestion to determine which minipreps contain the library plasmid. Alternatively, the E.coli transformations are plated onto LB amp plates containing X-Gal; transformants that contain the lacZ reporter plasmid form light blue colonies and are not picked. Restriction analysis on minipreps from the white colonies is used to distinguish between the library and bait plasmids. ______________________________________________________ Protocol 5. Isolating and classifying library plasmids. Several effective methods are available for isolating plasmids from yeast in amounts sufficient for E.coli transformation (e.g. the "smash and grab" method of Hoffman and Winston (51)). The method described below is quick, and yields plasmid DNA clean enough to transform E.coli efficiently and to work as a template for PCR. 1. Scrape a large mass of yeast from a plate and resuspend in 1 ml of TE in an eppendorf tube (the OD600 of this suspension should be between 2 and 5). Yeast from a fresh, 2 to 3 day old plate work best. The yeast can also be obtained from a 1 ml overnight liquid culture. 2. Spin briefly to pellet cells. Resuspend yeast in 0.5 ml S buffer (10 mM KPO4 pH 7.2, 10 mM EDTA, 50 mM 2-mercaptoethanol, 50 mg/ml zymolase). 3. Incubate at 37oC for 30 min. 4. Add 0.1 ml lysing solution (0.25 M Tris Cl pH7.5, 25 mM EDTA, 2.5% SDS). Vortex to mix. 5. Incubate at 65oC 30 min. 6. Add 166 ml 3 M KOAc. Chill on ice for 10 min. 7. Spin in an eppendorf centrifuge for 10 min. Pour supernate into a new tube. 8. Precipitate DNA by adding 0.8 ml cold ethanol. Incubate on ice for 10 min, spin for 10 min, and pour off supernate. 9. Wash pellet with 0.5 ml 70% ethanol and dry pellet. 10. Disolve the pellet in 40 ml sterile water. Use 1 or 2 ml of this crude yeast miniprep to transform E.coli by electroporation. Before transforming E.coli with the crude yeast minipreps it is often useful to determine which yeast minipreps contain the same library plasmid so that fewer need to be characterized. This can be done by restriction analysis of PCR products. 11. Set up a 0.5 ml tube for each yeast miniprep containing: ** 13 ml sterile water ** 2 ml 10X Taq polymerase buffer ** 2 ml dNTP mix (all 4 dNTPs at 2.5 mM each) ** 1 ml 5' primer (0.1 mg/ml) derived from the library vector fusion moiety sequence (see appendix). ** 1 ml 3' primer (0.1 mg/ml) derived from the ADH1 terminator sequence in the library vector (see appendix). ** 0.2 ml Taq polymerase (5 units/ml) 12. Add 1 ml of yeast miniprep DNA to each tube. 13. Incubate for 25 cycles of 92oC for 30 sec., 65oC for 2 min, 75oC for 30 sec. 14. Set up two tubes for each PCR reaction, one for AluI digestion and one for HaeIII digestion (these enzymes are recommended because they work well in the presence of the PCR reactants). Add 8 ml of the PCR reaction to each tube. Save the remainder of the PCR reaction for gel analysis of the full-length PCR product. 15. Add 1 ml of the appropriate 10X restriction enzyme buffer to the 8 ml. Add 1 ml of AluI to one tube and 1 ml of HaeIII to the other tube. Incubate at 37oC for 2 hours. 16. Analyze AluI digests, HaeIII digests, and the uncut PCR products on 1.5% agarose gels. It should be readily apparent from this analysis which cDNAs are identical. In some cases, only some of the restriction fragments will appear identical between two plasmids, suggesting that these plasmids contain different length cDNAs made from the same gene. Rescue library plasmids from yeast minipreps that give different restriction patterns. ______________________________________________________ ______________________________________________________ Protocol 6. Determining specificity of interactors. 1. Use the rescued library plasmid DNA (from a miniprep of transformed E. coli) to transform the original yeast strain containing the bait plasmid and the lacZ reporter plasmid. Additionally, use each library plasmid to transform one or more other strains containing different baits. Select transformants on Glu ura-his-trp- plates. As a control transform each different bait strain with the library vector without a cDNA insert. 2. Use sterile toothpicks or applicator sticks to pick three to four individual colonies from each transformation plate, and patch these to Glu ura-his-trp- master plates. Patch control transformants (library plasmid with no cDNA) onto each plate for side by side comparison. Grow for one or two days at 30oC. 3. Replica plate from the master plates to four plates: 1. Glu ura-his- trp- X-Gal, 2. Gal/Raf ura-his-trp- X-Gal, 3. Glu ura-his-trp-leu-, 4. Gal/Raf ura-his-trp-leu-. 4. Incubate plates at 30oC and examine after 1, 2, and 3 days. Figure 3 shows an example of the result obtained for three specific interactors. _____________________________________________________________ (Figure 3. Four plates showing specific interactors.) 6. Using a mating assay to verify specificity. We have recently developed a mating assay as an alternative way to test for interaction between a given activation-tagged protein and a panel of LexA fusion proteins (baits). This scheme takes advantage of the fact that haploid cells of the opposite mating type will fuse to form diploids when brought into contact with each other (52). In this mating assay, the activation-tagged protein is expressed in one yeast strain and the bait is expressed in a second strain of the opposite mating type. When the two strains are mixed on the same plate, they form diploids in which the bait and activation-tagged proteins have the opportunity to interact and activate the reporter genes. As before, interaction is measured as activation of the LexAop-LEU2 and LexAop-lacZ reporters. This technique can be used to check the specificity of the cDNA-encoded proteins isolated in the interaction trap to ensure that they interact with only the original bait and not with unrelated LexA fusions. It can also be used to examine interactions between a given set of activation- tagged proteins, such as those isolated in an interactor hunt, and a large panel of bait proteins without performing an unwieldy number of yeast transformations. Finally, it could be used to conduct interactor hunts; the library plasmids can be introduced in bulk into EGY48, transformants can be stored frozen, and then thawed and mated with a second strain that contains a bait plasmid. This would enable several separate interactor hunts with different baits to be done by performing a single large scale library transformation. The disadvantage of the mating assay is that the sensitivity of the reporters to activation by baits and activation-tagged proteins that interact with them is generally less in diploid cells relative to haploid cells. To minimize the effect of this difference, the most sensitive LEU2 reporter is used. As described in Protocol 7 and illustrated in Figure 4, the activation-tagged protein is expressed in EGY48 (mating type a) and the bait protein is expressed in a second strain, RFY206 (mating type a) which may also contains the lacZ reporter plasmid. The EGY48 derivatives are streaked onto plates lacking tryptophan to maintain selection for the library plasmid, and the RFY206 derivatives are streaked onto plates lacking uracil and histidine to maintain selection for the URA3 lacZ plasmid and the HIS3 bait plasmid. The two strains are mated by applying them to the same replica velvet or filter and lifting their "print" with a YPD plate. The YPD plate is incubated overnight at 30oC to promote mating and then replica plated to the same indicator plates used in Protocol 4 steps 7 and 8. In the example shown in Figure 4 the lacZ reporter is not used. The his-trp-plates will contain only diploids: TRP1 is provided by the strain that contains the library plasmid while HIS3 is provided by the strain that contains the bait plasmid. Only the diploids that contain interacting pairs of activation-tagged protein and bait will grow on the plates lacking leucine but containing galactose. (Figure 4a. Mating assay cartoon.) (Figure 4b. Mating assay result.) ______________________________________________________ Protocol 7. Mating assay. 1. Introduce TRP1 library plasmids into yeast strain EGY48 and select transformants on Glu trp- plates. As a control, transform EGY48 with a library plasmid that has no cDNA insert. 2. Introduce HIS3 bait plasmids into yeast strain RFY206 (or other ura3 his3 trp1 leu2 MATa strain) along with a URA3 lacZ reporter and select transformants on Glu ura-his- plates. Note: in the example shown in Figure 4 the URA3 lacZ reporter is not used. 3. Use sterile toothpicks or applicator sticks to streak individual EGY48 transformants onto standard 100 mm Glu trp- plates in parallel lines, 6 or 7 to a plate (Figure 4). Include at least one streak of the transformants with the control plasmid (no cDNA). Likewise, streak individual RFY206 transformants onto Glu ura-his- plates in parallel lines, 6 or 7 to a plate. Incubate at 30oC until heavy growth. The lines of yeast should be at least 2 mm wide. 4. Onto the same replica filter or velvet, print the EGY48 derivatives and the RFY206 derivatives so that the streaks from the two plates are perpendicular to each other. 5. Lift the print of the two strains from the filter or velvet with a YPD plate. Incubate the YPD plate at 30oC overnight. Diploids will form where the two strains intersect. One strain may grow more rapidly than the other during this time but this does not hinder formation of diploids in the intersections. 6. Replica from the YPD plate to the following plates: 1. Glu ura-his- trp- X-Gal, 2. Gal/Raf ura-his-trp- X-Gal, 3. Glu ura-his-trp-leu-, 4. Gal/Raf ura-his-trp-leu-. Incubate at 30oC and examine after 1, 2, and 3 days. Only diploids will grow on the X-Gal plates and only interactors will grow on galactose plates lacking leucine (for example, see Figure 4). ______________________________________________________ 7. Expected results. The two most critical parameters that determine whether an interactor hunt will succeed are the quality of the library and the nature of the bait. If the library contains cDNAs that encode proteins that interact with the bait in the interaction trap, the bait becomes the most critical parameter. The bait can affect the outcome of an interactor hunt in two ways. First, the number of nonspecific cDNA- encoded interactors obtained appears to depend on the bait. For example, use of some baits can result in isolation of activation-tagged cDNA proteins that seem to interact with many other LexA fusions when the specificity determination is performed. It is unlikely that these cDNA-encoded proteins interact with the LexA moiety or with the reporters or transcription machinery directly because if they did, they would be expected to arise in all interactor hunts. Rather, these cDNA-encoded proteins may simply be "sticky", for example, because they contain patches of hydrophobic or charged residues that interact with corresponding regions on many baits. Second, the number of spurious Leu+ yeast that arise independently of the library plasmid differs from one bait to another. For example, when some selection strains are transformed with library DNA and plated onto media lacking leucine, hundreds of galactose-independent Leu+ colonies grow. This phenomenon is observed more frequently with baits that activate a low level of transcription than with transcriptionally inert baits, perhaps because a low level of activation is readily enhanced in a fraction of cells in a population. For example, a bait may gain the ability to activate the reporters in some cells if there is an increase in the copy number of the bait plasmid, which could result from natural variation in plasmid copy number or from a mutation in the plasmid or a yeast gene that affects plasmid copy number. This problem can be best addressed by reducing the activation potential of a bait as described in section 4.2. It is worth noting that, in our experience, these problems with baits are usually surmountable, and that most proteins can eventually be used successfully in interactor hunts. Conclusion We have described detailed procedures for isolating proteins using the interaction trap. These procedures provide a relatively rapid, simple, and robust method for isolating proteins that interact with known proteins, and for measuring protein-protein interactions inside living cells. Acknowledgment We are very grateful to the numerous current and past lab members, and to the many lab visitors whose efforts contributed to the development of this interaction technology. We thank Tod Gulick, Pierre Colas, and Andrew Mendelsohn for critcal readings of the manuscript. RLF was supported by a postdoctoral fellowship from the NIH. RB was supported by the Pew Scholar's program. Work was supported by the HFSP. Appendix Figure of plasmids. pEG202 pLR1d1 (pSH18-34, pJK103, pJK101) pJG4-5 Sequencing and PCR primers for pEG202 and pJG4-5. For sequencing inserts in the polylinker of pEG202 primers can be derived from LexA coding sequences. The primer shown below, LEX1, is derived from LexA coding sequences 40 bp upstream of the EcoRI site in the polylinker (note: all eukaryotic LexA expression plasmids lack the EcoRI site found in the wild-type LexA coding region (18)). For sequencing from the 5' end of cDNA inserts in pJG4- 5, BCO1 can be used. It is derived from the coding sequence for the B42 activation domain 70 bp upstream of the EcoRI site. For sequencing from the 3' end of the cDNA insert in pJG4-5, BCO2 can be used. It is derived from the sequence of the ADH1 terminator approximately 40 bp downstream of the XhoI site. BCO1 and BCO2 can be used for PCR amplification of the cDNA insert as described in Protocol 5. LEX1 5' CGT CAG CAG AGC TTC ACC ATT G 3' BCO1 5' CCA GCC TCT TGC TGA GTG GAG ATG 3' BCO2 5' GAC AAG CCG ACA ACC TTG ATT GGA G 3' Media recipes. Dropout media. Dropout media, also known as complete minimal (CM) dropout media, contains a nitrogen base, a mixture of nutrients shown in Table 1 with one or a few left out (dropped out), and a carbon source (usually a sugar). It is convenient to make a dropout powder corresponding to each of the dropout media that will be used, e.g. for media lacking tryptophan (trp-), the dropout powder used would contain all of the nutrients listed in Table 1 except for tryptophan. Three separate stocks of the carbon sources (20% galactose, 20% glucose, 20% raffinose) should be made and filter sterilized. As needed, galactose (Gal) and glucose (Glu) are each added to 2 % final concentration, raffinose (Raf) is added to 1 % final concentration. Dropout plates For 1 liter Mix in 850 ml deionized H2O ** 6.7 g yeast nitrogen base (YNB) without amino acids (Difco) ** 2 g Dropout powder lacking the appropriate nutrient(s) (Table 1) ** one pellet of NaOH (~ 0.1 g) ** 20 g agar (Difco Bacto-agar) Autoclave Add the appropriate carbon source from sterile stocks. For Gal/Raf plates add galactose to 2% and raffinose to 1% final concentrations; for Glu plates add glucose to 2 % final concentration. Make liquid dropout media the same way as dropout plates, leaving out the agar and NaOH pellet. Table 1. Dropout powder. Nutrient Amt in dropout powder (g)a Final conc. in media (ug/ml) adenine 2.5 40 L-arginine (HCl) 1.2 20 L-aspartic acid 6.0 100 L-glutamic acid (monosodium) 6.0 100 L-histidine (his) 1.2 20 L-isoleucine 1.8 30 L-leucine (leu) 3.6 60 L-lysine 1.8 30 L-methionine 1.2 20 L-phenylalanine 3.0 50 L-serine 22.5 375 L-threonine 12.0 200 L-tryptophan (trp) 2.4 40 L-tyrosine 1.8 30 L-valine 9.0 150 uracil (ura) 1.2 20 a Combine all but the appropriate nutrients, e.g. if making media lacking tryptophan (trp-) leave out tryptophan. Grind in a clean dry mortar and pestle until homogeneous. Store at room temperature. YPD (also known as YEPD) plates. For 1 litre Mix in 900 ml deionized H2O ** 10 g yeast extract (Difco Bacto-yeast extract) ** 20 g peptone (Difco Bacto-peptone ** one pellet NaOH (~0.1 g) ** 20 g agar (Difco Bacto-agar) Autoclave Add 100 ml sterile 20% glucose X-Gal plates. For 1 litre Mix in 800 ml deionized H2O ** 6.7 g YNB without amino acids ** 1.5 g dropout powder ** 20 g agar (Difco Bacto-agar) Autoclave Immediately add the appropriate sterile 20% carbon source(s) Allow to cool to 65oC ** add 100 ml of 10X BU Salts (see below) ** add 4 ml of 20 mg/ml X-Gal dissolved in DMF dimethyl formamide (stored at -20oC). Note: adding salts while media is too hot causes the salts to precipitate. Also, X-Gal is thermolabile and will be destroyed if added to hot media. 10X BU Salts For 1 liter ** 70 g Na2HPO4 7H2O ** 30 g NaH2PO4 Adjust pH to 7.0 Autoclave, store at room temp. Figure Legends Figure 1. The interaction trap. a. Glucose. The LexA fusion protein (bait) is made and binds to LexA operators (black box) upstream of the two reporter genes, LEU2 and lacZ. The bait does not activate transcription of the reporters. The activation-tagged cDNA-encoded protein is not expressed because the GAL1 promoter on the library plasmid is repressed in the presence of glucose. The cell does not form a colony on a medium lacking leucine and forms a white colony on an X-Gal plate. b. Galactose. Here, galactose induces expression of an activation- tagged cDNA-encoded protein that does not interact with the bait. The cell does not grow on a medium lacking leucine and forms a white colony on X-Gal plates. c. Galactose. Here, galactose induces expression of an activation- tagged cDNA-encoded protein that interacts with the bait. The activation domain activates transcription of LEU2 and lacZ. The cell forms a colony on a medium lacking leucine and forms a colony that turns blue on an X-Gal plate. Figure 3. Specificity test. Specificity of Drosophila Cdc2 kinase interactors (Cdis; Finley and Brent, in preparation). Four replica plates made from a glucose master plate (not shown) that contained EGY48 derivatives with different bait plasmids and library plasmids. All strains contained the lacZ reporter plasmid pJK103, and one of four bait plasmids directing the synthesis of LexA fused to Drosophila proteins: Cdc2 kinase (Cdc2), a Cdc2 kinase analog (Cdc2c), a derivative of the homeodomain protein Bicoid (BcdDC), or the helix-loop-helix protein Hairy (Hairy). Each bait strain was transformed with the library vector pJG4-5 (v), or with pJG4-5 that contained cDNAs that encode Cdc2 kinase interactors: Cdi3 (3), Cdi2 (2), and Cdi7 (7). Four individual colonies from each transformation were patched onto a glucose master plate which was then replica plated to two ura-his- trp-leu- plates (leu-), and two ura-his-trp- X-Gal plates (XGal). The plates on the left have glucose and the plates on the right have galactose and raffinose. Interaction is detected by growth of strains on the galactose leu- plate; e.g., Cdc2 interacts with Cdi3, Cdi2 and Cdi7; Cdc2c interacts with Cdi3 and Cdi2. The strength of the interactions is suggested by the level of activation of lacZ as indicated on the X-Gal plate; e.g., the interaction between Cdc2 and Cdi2 activates lacZ strongly, the interaction between Cdc2 and Cdi7 activates lacZ very weakly, and the other interactions activate lacZ moderately. Figure 4. The mating assay for specificity. a. A typical mating assay (Finley and Brent, in preparation). The his- glucose plate (top left) contains four RFY206 derivatives streaked vertically. Each derivative contains a different HIS3+ bait plasmid. The trp- glucose plate (top right) contains seven EGY48 derivatives streaked horizontally. Each derivative contains a different TRP1 library plasmid. The RFY206 derivatives are MAT a, HIS3, trp1-, and Leu-. The EGY48 derivatives are MAT a TRP1, his3-, and Leu-. The two plates are pressed to the same replica velvet or filter and the replica is lifted with a YPD plate (center). During overnight incubation at 30oC the two strains grow. At the intersections on the YPD plate the two strains mate and form His+ Trp+ diploids. The YPD plate is then replica plated to three plates: 1. a his-trp- galactose plate (with raffinose), on which diploids grow, but neither haploid parent grows; 2. a his-trp-leu- glucose plate, on which the activation- tagged cDNA encoded proteins are not expressed, LEU2 is not transcribed, and no strains grow; 3. a his-trp-leu- galactose plate (with raffinose), on which activation-tagged cDNA encoded proteins are expressed, interact with the baits, activate the LEU2 gene, and allow growth. b. Results of a mating assay. The horizontally streaked strains are RFY206 derivatives expressing LexA fusions to Drosophila Cdc2 kinase (Cdc2); the Cdc2 kinase analog, Cdc2c (Cdc2c); a Bicoid derivative (BcdDC); Hairy (Hairy); the budding yeast Cdc2 kinase homolog, Cdc28 (Cdc28), and two proteins isolated from a hunt for Cdc2c interactors (Cdi5 and Cdi11) (Finley and Brent, in preparation). The vertically streaked strains are EGY48 derivatives that contain the pJG4-5 library vector without a cDNA insert (v) or with Cdc2 kinase interactor cDNAs, CDI2 (2), CDI3, (3), and CDI7 (7). Appendix Figure legends. pEG202. pEG202 (11) is a yeast - E.coli shuttle vector that contains a yeast expression cassette that includes the promoter from the yeast ADH1 gene (PADH1), sequences that encode amino acids 1 to 202 of the bacterial repressor protein LexA, a polylinker, and the transcription terminator sequences from the yeast ADH1 gene (TADH1). It also contains a E.coli origin of replication (pBR ori), the ampicillin resistance gene (AmpR), a yeast selectable marker gene (HIS3), and a yeast origin of replication (2 mm ori). pEG202 confers upon a his3- yeast strain the ability to grow in the absence of histidine and directs the constitutive expression of LexA (fused to approximately 17 amino acids encoded by the polylinker). Protein coding sequences can be inserted in-frame with LexA into the unique restriction sites shown. PJG4-5. pJG4-5 (11) is a yeast - E.coli shuttle vector that contains a yeast expression cassette that includes the promoter from the yeast GAL1 gene (PGAL1), followed by sequences that encode the 106 amino acid fusion moiety or activation tag, and the transcription terminator sequences from the yeast ADH1 gene (TADH1). cDNAs or other protein coding sequences can be inserted into the unique EcoRI and XhoI sites so that encoded proteins are expressed with the fusion moiety at their amino terminus. The fusion moiety includes the nuclear localization signal from the SV40 virus large T antigen (PPKKKRKVA; (53)), the B42 transcription activation domain (4), and the hemagglutinin (HA) epitope tag (YPYDVPDYA; (54)). The plasmid also contains an E.coli origin of replication (pUC ori), the ampicillin resistance gene (AmpR), a yeast selectable marker gene (TRP1), and a yeast origin of replication (2 mm ori). lacZ reporters. The lacZ reporters are derived from a plasmid that contains the wild- type GAL1 promoter fused to lacZ (19). Reporters for measuring activation are derived from pLR1D1, in which the GAL1 upstream activation sequences (UASg) have been deleted (33). Various numbers and types of LexA operators have been inserted in place of UASg to create lacZ reporters with different sensitivities. Two examples are shown: pJK103 contains a single high affinity overlapping type colE1 LexA operator which binds two LexA dimers; pSH18-34 contains four of these colE1 operators and so is much more sensitive than pJK103 to activation by LexA fusions and activation- tagged proteins that interact with them. Derivatives with two, three, or more than four colE1 operators have also been made (S. Hanes, personal communication; (28)). A reporter less sensitive than pJK103, pRB1840, contains one lower affinity LexA operator (3). pJK101 is used to measure repression by LexA fusions (18; 55) It contains most of UASg and one colE1 operator between UASg and the GAL1 TATA . All of the lacZ reporters also contain a E.coli origin of replication (pBR ori), the ampicillin resistance gene (AmpR), a yeast selectable marker gene (URA3), and a yeast origin of replication (2 mm ori). References 1. Keegan, L., Gill, G. and Ptashne, M. (1986). Science, 231, 699-704. 2. Hope, I. A. and Struhl, K. (1986). Cell, 46, 885-894. 3. Brent, R. and Ptashne, M. (1985). Cell, 43, 729-736. 4. Ma, J. and Ptashne, M. (1987). Cell, 51, 113-119. 5. Sadowski, I., Ma, J., Triezenberg, S. and Ptashne, M. (1988). Nature, 335, 563-564. 6. Chasman, D. I., Leatherwood, J., Carey, M., Ptashne, M. and Kornberg, R. D. (1989). Mol. Cell. Biol., 9, 4746-4749. 7. Ma, J. and Ptashne, M. (1988). Cell, 55, 443-446. 8. Fields, S. and Song, O. (1989). Nature, 340, 245-246. 9. Chien, C.-T., Bartel, P. L., Sternglanz, R. and Fields, S. (1991). Proc. Natl. Acad. Sci. USA, 88, 9578-9582. 10. Durfee, T., Becherer, K., Chen, P.-L., Yeh, S.-H., Yang, Y., Kilburn, A. E., Lee, W.-H. and Elledge, S. J. (1993). Genes and Dev., 7, 555-569. 11. Gyuris, J., Golemis, E., Chertkov, H. and Brent, R. (1993). Cell, 75, 791-803. 12. Vojtek, A. B., Hollenberg, S. M. and Cooper, J. A. (1993). Cell, 74, 205-214. 13. Dalton, S. and Treisman, R. (1992). Cell, 68, 597-612. 14. Silver, P. A., Keegan, L. and Ptashne, M. (1984). Proc. Natl. Acad. Sci. U.S.A., 81, 5951-5955. 15. Bram, R. J. and Kornberg, R. D. (1985). Proc. Natl. Acad. Sci. U.S.A., 82, 43-47. 16. Giniger, E., Varnum, S. M. and Ptashne, M. (1985). Cell, 40, 767- 774. 17. Silver, P. A., Brent, R. and Ptashne, M. (1986). Mol. Cell. Biol., 6, 4763-4766. 18. Brent, R. and Ptashne, M. (1984). Nature, 312, 612-615. 19. Yocum, R. R., Hanley, S., West, R. J. and Ptashne, M. (1984). Mol. Cell. Biol., 4, 1985-1988. 20. Guarente, L. and Ptashne, M. (1981). Proc. Natl. Acad. Sci. USA, 78, 2199-2203. 21. Guthrie, C. and Fink, G. R. (1991). Guide to yeast genetics and molecular biology, 194, Academic Press, Inc., Boston . 22. Ausubel, F. M., Brent, R., Kingston, R. E., Morre, D., Seidman, J. G. and Struhl, K. (1992). Current protocols in molecular biology, Greene and Wiley-interscience, New York . 23. Rothstein, R. (1985). DNA cloning: A practical approach. (D. M. Glover), II, 45-66. IRL Press, Ltd, Oxford. 24. Golemis, E. and Brent, R. (1992). Mol. Cell. Biol., 12, 3006-3014. 25. Zervos, A. S., Gyuris, J. and Brent, R. (1993). Cell, 72, 223-232. 26. Lech, K., Anderson, K. and Brent, R. (1988). Cell, 52, 179-184. 27. Samson, M.-L., Jakson-Grusby, L. and Brent, R. (1989). Cell, 57, 1045-1052. 28. Hanes, S. and Brent, R. (1989). Cell, 57, 1275-1283. 29. Kamens, J., Richardson, P., Mosialos, G., Brent, R. and Gilmore, T. (1990). Mol. Cell. Biol., 10, 2840-2847. 30. Ebina, B. A., Takahara, Y., Kishi, F., Nakazawa, A. and Brent, R. (1983). J. Biol. Chem., 258, 13258-13261. 31. Gill, G. and Ptashne, G. (1987). Cell, 51, 121-126. 32. Broach, J. R. (1981). The molecular biology of the yeast Saccharomyces: Life cycle and inheritance (J. N. Strathern, E. W. Jones and J. R. Broach), 445-470. Cold Sping Harbor Laboratory, Cold Spring Harbor. 33. West Jr., R. W., Yocum, R. R. and Ptashne, M. (1984). Mol. Cell. Biol., 4, 2467-2478. 34. Manivasakam, P. and Schiestl, R. H. (1993). Nuc. Acids Res., 21, 4414-4415. 35. Becker, D. M. and Guarente, L. (1991). Methods in Enzymology, 194, 182-187. 36. Gietz, D., St. Jean, A., Woods, R. A. and Schiestl, R. H. (1992). Nuc. Acids Res., 20, 1425. 37. Brent, R. (1985). Cell, 42, 3-4. 38. Finley Jr., R. L., Chen, S., Ma, J., Byrne, P. and West Jr., R. W. (1990). Mol. Cell. Biol., 10, 5663-5670. 39. Miller, J. (1972). Experiments in molecular genetics, Cold Spring Harbor Laboratory, Cold Spring Harbor, N.Y. 40. Finley Jr., R. L. and R.W., W. J. (1989). Mol. Cell. Biol., 9, 4282- 4290. 41. Laemmli, U. K. (1970). Nature, 227, 680-685. 42. Harlow, E. and Lane, D. (1988). Antibodies: A laboratory manual 471-510. Cold Spring Harbor Laboratory, Cold Spring Harbor, N.Y. 43. Gill, G. and Ptashne, M. (1988). Nature, 334, 721-724. 44. Berger, S. L., Pina, B., Silverman, N., Marcus, G. A., Agapite, J., Regier, J. L., Triezenberg, S. J. and Guarente, L. (1992). Cell, 70, 251- 265. 45. Gubler, U. and Hoffman, B. J. (1983). Gene, 25, 263-269. 46. Huse, W. D. and Hansen, C. (1988). Stratagene stratagies, 1, 1-3. 47. Schiestl, R. H. and Gietz, R. D. (1989). Curr. Gen., 16, 339-346. 48. Hill, J., Ian, K. A. and Griffiths, D. E. (1992). Nuc. Acids Res., 19, 5791. 49. Breeden, L. and Nasmyth, K. (1985). 50, 643. Cold Spring Harbor, NY. 50. Struhl, K., Stinchomb, D. T., Scherer, S. and Davis, R. W. (1979). Proc. Natl. Acad. Sci., 76, 1035-1039. 51. Hoffman, C. S. and Winston, F. (1987). Gene, 57, 267-272. 52. Herskowitz, I. (1988). Microbiol. Rev., 52, 536-553. 53. Kalderon, D., Roberts, B. L., Richardson, W. D. and Smith, A. E. (1984). Cell, 39, 499-509. 54. Green, N., Alexander, H., Olson, A., Alexander, S., Shinick, T. M., Sutcliffe, J. G. and Lerner, R. A. (1982). Cell, 28, 477-487. 55. Kamens, J. and Brent, R. (1991). The New Biologist, 3, 1005-1013.