Molecular Microbiology (2008) 67(4), 719–728 䊏

doi:10.1111/j.1365-2958.2007.06077.x First published online 2 January 2008

A phage display system designed to detect and study protein–protein interactions Catherine L. Bair,1 Amos Oppenheim,1,2† Andrei Trostel,1 Gali Prag3 and Sankar Adhya1* 1 Laboratory of Molecular Biology, Center for Cancer Research, National Cancer Institute, National Institutes of Health, Bethesda, MD 20892, USA. 2 Department of Molecular Genetics and Biotechnology, Hebrew University-Hadassah Medical School, Jerusalem 91120, Israel. 3 Department of Biochemistry, George S. Wise Faculty of Life Sciences Tel Aviv University Tel Aviv, Israel.

Summary Analysing protein–protein interactions is critical in proteomics and drug discovery. The usage of 2-Hybrid (2l) systems is limited to an in vivo environment. We describe a bacteriophage 2-Hybrid system for studying protein interactions in vitro. Bait and prey are displayed as fusions to the surface of phage l that are marked with different selectable drug-resistant markers. An interaction of phages in vitro through displayed proteins allows bacterial infection by two phages resulting in double drugresistant bacterial colonies at very low multiplicity of infections. We demonstrate interaction of the protein sorting signal Ubiquitin with the Vps9-CUE, a Ubiquitin binding domain, and by the interaction of (Gly-Glu)4 and (Gly-Arg)4 peptides. Interruptions of the phage interactions by non-fused (free) bait or prey molecules show how robust and unique our approach is. We also demonstrate the use of Ubiquitin and CUE display phages to find binding partners in a l-display library. The unique usefulness to 2l is also described.

Introduction The identification and characterization of protein–protein interactions are important as most proteins function in complexes with other proteins. Yeast 2-Hybrid, pull down, protein microarray, and immunoaffinity chromatography assays are commonly used for probing protein–protein interactions. The realization of numerous open reading Accepted 1 December, 2007. *For correspondence. E-mail adhyas@ mail.nih.gov; Tel. (+1) 301 496 2495; Fax (+1) 301 402 1344. †In memory of the deceased.

Journal compilation © 2008 Blackwell Publishing Ltd No claim to original US government works

frames (ORFs) of unknown function across many generi demands a protein association assay that is specific, sensitive and flexible if one hopes to discover the biochemical partners of novel gene products, or attempt to diagram and understand uncharacterized protein networks. The yeast 2-Hybrid (Fields and Song, 1989; Auerbach et al., 2002) (and the more recent 1-Hybrid and 3-Hybrid iterations) (Li and Herskowitz, 1993; Zhang and Lautar, 1996) and bacterial 2-Hybrid (Karimova et al., 1998; Joung et al., 2000; Zhu et al., 2000; Dove and Hochschild, 2004) display systems have been valuable tools for studying protein associations. In principle, these methods use a tagged protein (bait) that finds a binding partner (prey) in a library of other proteins within an in vivo environment by screening or selection. The objective of a display system is to identify the highest number of binding partners with the lowest possible background, and be able to construct high titer libraries to increase the chance of each protein being represented. The phage display technique has historically been limited to bio-panning (Zucconi et al., 2001) against immobilized targets, and has shown to be both sensitive and highly advantageous for the enrichment of specific binding partners from a library (Kay et al., 1996; Cicchini et al., 2002). Recently, bacteriophage display has become a tool of choice to study functional genomics as it continually proves to be a viable alternative to the lower titer yeast display system (~106) (Sternberg and Hoess, 1995; Auerbach et al., 2002; Hoess, 2002; Gupta et al., 2003; Kirsch et al., 2005) because one can create very high titer libraries (~109) using lambda (a minimum of 108 clones with an average insert size of 3 kb is necessary to ensure representation of a rare gene) (Hagen et al., 1988). Compared with the phage M13 display system, lambda-display shows higher tolerance for larger proteins (Gupta et al., 2003). We developed a bacteriophage l-based version of the 2-Hybrid system, herein named 2l, that we propose is better suited than other hybrid systems in several ways for studying protein interactions.

Results Principle of protein association assay by l display Bacteriophage l upon infection of its host Escherichia coli shows one of two lifestyles. Depending upon bacterial

720 C. L. Bair et al. 䊏

Fig. 1. The principle of the 2l system for protein–protein interactions. Displayed is the association of two proteins on two different phages, and subsequent dual infection of a cell resulting in a double antibiotic resistant lysogen.

growth conditions, l either shows lytic growth by producing progeny phage particles and lysing the host, or becomes a prophage within the host by repressing most of its lytic genes and integrating its genome into the host chromosome. The prophage state can be induced to lytic growth by DNA damaging agents. At a multiplicity of infection (moi) = 1, l predominantly shows lytic growth (Kourilsky, 1973; Kourilsky and Knapp, 1974; Oppenheim et al., 2005). However, cells more often become lysogens if the moi is 2 or more. The chance of two phage particles infecting the same cell at an average moi much less than 1, for example 1 ¥ 10-3, is one in a million. We argue that if at such a low moi if two protein molecules displayed on two different phage particles interact with each other before infecting the host, lysogeny could ensue because the localized moi would effectively be two for that host cell (Fig. 1). It is this local concentration of two or more

phages resulting from interactions of displayed proteins that forms the basis for the 2l protein association assays. An E. coli cell lysogenic for one or both of the display phages can be induced to rescue the prophage(s) allowing for direct sequencing of the genes, and confirmation and identification of the interacting proteins. Therefore, if a library of genes is expressed on one or both of the phages a direct link can be made between the proteins through their phage-encoded DNA. The construction of phages displaying proteins is described in Experimental procedures and diagrammed in Fig. 2. We facilitated the isolation of double lysogens by cloning two different selectable antibiotic resistant markers, cml R and kan R, in two l display vectors, henceforth named lc and lk respectively (Fig. 2). The resistance genes were introduced at identical sites in the phage genome between the genes R and right cos end to eliminate the possibility of obtaining l recombinants carrying both resistance genes. The double lysogens of the two l display phage types can be selected directly on bacterial agar plates containing chloramphenicol and kanamycin. In order to simplify the rescue of the phage from lysogens, the vector phage carries a temperature-sensitive cI repressor mutant in its imm21 allele. This allows lytic development upon shifting a lysogenic log phase culture from 32°C to 42°C and thus high yield of phage particles can easily be obtained. Association assays by the 2l system of known or expected binding partners When bacterial cells were mixed with either a lk or lc phage lysate, as anticipated, neither lk nor lc was able to form significant numbers of antibiotic resistant lysogens (CmlR or KanR) at an average moi of less than 0.1 (data not shown). Only at a moi of 2 or above, incidental Fig. 2. Diagrammatic representation of cloning ORFs at the multiple cloning sites (mcs) of the plasmid pVCDcDL3 and the Cre-lox mediated recombination for construction of l display phage as described in Gupta et al. (2003).

Journal compilation © 2008 Blackwell Publishing Ltd, Molecular Microbiology, 67, 719–728 No claim to original US government works

A phage display system 721

coinfection by the same phage and thereby lysogen formation becomes significant. Similar results were found when the phage vectors lk and lc were mixed and assayed for double antibiotic resistant lysogen formation. The frequency of CmlR/KanR lysogens was 0 from a total moi (e.g. moi of lk + moi of lc versus 1 ¥ 108 cells) of 2 ¥ 10-6-2 ¥ 10-3, less than 5 at a moi of 2 ¥ 10-2, and about 10 at a moi of 2 ¥ 10-1 (Fig. 3A). At a total moi = 2 or more, when the phages outnumber the cells, the vector phages were able to form lysogens at a significant frequency. We obtained an average of 105 CmlR/KanR colonies at a moi = 2. In the negative control there were no double antibiotic resistant lysogens formed when cells were infected with only lk or lc. We assayed the ability of the 2l system to demonstrate protein–protein interactions by studying protein and peptide pairs that interact through hydrophobic or hydrophilic groups. The well-described protein complex of the CUE (coupling of Ubiquitin conjugation to ER degradation) domain with the signal sorting protein Ubiquitin was chosen to investigate if the system can display protein domains for fruitful interactions (Shih et al., 2003). The CUE and UBA (Ubiquitin associated) domains are part of a large superfamily of three-helical Ubiquitin binding domains. At the interface of the complex Vps9CUE:Ubiquitin, 39 out of the 43 residues involved in the association participate in hydrophobic bonding (Prag et al., 2003; Shih et al., 2003). We cloned the DNA segments encoding the Ubiquitin and CUE peptides into the l D-display vectors, as described in Experimental procedures. When used individually for cell infection, each of the four lkD-CUE, lcDCUE, lkD-Ubiquitin and lcD-Ubiquitin fusion display phages was as viable in cell infection by plaque formation and single antibiotic resistance lysogen formation as the vector phages. When either the lcD-CUE:lkD-Ubiquitin or the lkD-CUE:lcD-Ubiquitin combination was used for coinfection of cells and selection for CmlR/KanR lysogens, lysogens were formed with an efficiency higher than that for vector phages. We obtained an average of about 5 CmlR/KanR lysogens at a moi of 2 ¥ 10-4 (Fig. 3B). The frequency of lysogen formation increased gradually to a plateau at a moi > 1. The increase in lysogen formation at a moi < 1 observed following infection by both display phages is likely due to the protein associations that join the two phages making both phages available for coinfection and establishment of lysogeny. Similar results were obtained with the corresponding lkD-CUE:lcD-Ubiquitin pair (data not shown). There were no double antibiotic resistant lysogens formed when cells were infected with lcD-CUE, lkD-Ubiquitin (Fig. 3B), lkD-CUE or lcDUbiquitin phage (data now shown). To investigate if the 2l system can also be used to study hydrophilic interactions, we designed two peptides con-

taining a leader sequence followed by a repeating pair of charged amino acids. One is a 17-amino-acid-long aptamer containing the repeating (Gly-Glu)4 sequence, designated as D-Acid. The other contains the repeating sequence (Gly-Arg)4, designated henceforth as D-Base. It is assumed that these peptides when displayed on phages will associate through electrostatic attractions, aided by the high concentration of negative and positive charges on their respective virion heads. The DNA sequences encoding the peptides were cloned into the lk and lcD-display vectors as in Experimental procedures. The efficiency of single antibiotic gene-resistant lysogen formation upon infection by lkD-Acid, lcD-Acid, lkD-Base or lcD-Base fusion display phage was comparable to that of the vector phages lk or lc. In contrast, there was a striking increase in frequency of lysogeny by lcD-Acid and lkD-Base mixed infection at a moi < 1 (Fig. 3C). Compared to the very low efficiency of lysogeny by the vector phages, we obtained an average of 7 CmlR/KanR lysogens at a moi of 2 ¥ 10-4. The frequency of lysogen formation increased with increasing moi and reached a plateau when the moi was much greater than 1. Identical results were obtained when cells were infected with the lkD-Acid and lcD-Base phages (data not shown). Again, there were no lysogens formed when cells were infected with any one of the four acid or base displaying phage, nor were there any significant number of double resistant lysogens observed at low moi when the following combinations of phages were used: lcD-Acid + lk, lc + lkD-Base or lcD-CUE + lk (data now shown). Specificity of l display phage interactions and determination of affinity We performed competition assays to verify that the observed binding of the displayed CUE and Ubiquitin proteins or the Acid and Base aptamers are specific, by testing if the binding partners could be titrated (detached) with one of the two displayed molecules in a given interacting pair. For this, display phages were mixed to give a total moi of 2.5 ¥ 10-4 in the presence of increasing concentrations of the free competitor molecule, mixed with cells, and then plated for lysogens on double antibiotic plates. We performed inhibition assays on the lcDCUE:lkD-Ubiquitin interaction using free Ubiquitin, CUE, or CUEM419D. Both free CUE and free Ubiquitin inhibited interactions between lcD-CUE and lkD-Ubiquitin phages, as was judged by more or less linear reduction of lysogen formation (Fig. 4A). The concentration at which free peptide inhibited lysogen formation by approximately 50% (apparent IC50) was 2 nM for CUE, and 20 nM for Ubiquitin. It is important to note that this apparent IC50 is not directly comparable to the IC50 calculated in vitro for association of the free peptides because of the multivalency of

Journal compilation © 2008 Blackwell Publishing Ltd, Molecular Microbiology, 67, 719–728 No claim to original US government works

722 C. L. Bair et al. 䊏

Fig. 3. Formation CmlR/KanR lysogens at different moi of premixed lk and lc vector or display phages. A. lk and lc vector phages. B. lcD-CUE and lkD-Ubiquitin. C. lcD-Acid and lkD-Base. For details see Experimental procedures. The addition of 1 ¥ 10x moi of phage #1 with 1 ¥ 10x moi of phage #2 result in a total moi of 2 ¥ 10x. The graphs are shown in log scale. The moi of 2 and 3 on the x-axis are not in scale with those below moi of 1.

Journal compilation © 2008 Blackwell Publishing Ltd, Molecular Microbiology, 67, 719–728 No claim to original US government works

A phage display system 723

Fig. 4. Inhibition assays of l display phase interactions. A. Inhibition of lcD-CUE and lkD-Ubiquitin at different inhibitor concentrations. Fusion display used to coinfect cells were at a moi of 0.04. Ubiquitin (diamond), CUE (square), mutant CUEM419D (triangle), acid aptamer (asterisk). The point (0.0) in both A and B represents the number of CmlR/KanR lysogens in the absence of inhibitors for the given moi. B. Inhibition of lcD-Acid and lkD-Base pair by free Acid and Base aptamers at different concentrations. Fusion display phage used to coinfect cells were at an moi of 0.0025. Acid aptamer (diamond), base aptamer (square), mutant CUE protein (triangle).

the D-fusion display peptides, steric hinderences between displayed proteins and the dynamics of displayed proteins binding and releasing on the virion head. In vitro, the CUE domain binds to Ubiquitin with a KD = 1.2 mM (Shih et al., 2003). The severe CUEM419D mutant (Prag et al., 2003), which was designed to decrease the affinity of CUE to Ubiquitin, failed to show inhibition of lcD-CUE and lkDUbiquitin phages, even when used at a concentration of 1 mM. The inhibitory effects of free CUE and Ubiquitin on the lcD-CUE and lkD-Ubiquitin interactions that result in the loss of phage–phage association were comparable to the relative inhibitory strength calculated for free proteins. The inhibition by the wild type CUE or Ubiquitin and the lack of inhibition by the non-binding CUEM419D derivative validate both the specificity of the interaction and retention of ‘native’ conformation of the fusion display proteins in the

display phages. The lkD-CUE and lcD-Ubiquitin pair was not inhibited by the acidic (Fig. 4A) or basic aptamers (data not shown). The interactions of lcD-Acid:lkD-Base were titrated with the free acidic peptide, giving an apparent IC50 10 nM, as shown (Fig. 4B). Similar results were obtained when the basic free peptide was used as a competitor. No crossinhibition was observed when the competition was carried out in the presence of non-specific peptides or proteins such as Ubiquitin, CUE or CUEM419D (see below). Arginine or glutamic acid inhibited the interactions at an apparent IC50 4.6 mM and 0.679 mM respectively (data not shown). We tested the ability of the binding partners to ‘find’ each other in the presence of excess non-specific phages (Table 1). No interference was observed in the lcDCUE:lkD-Ubiquitin interaction (at a moi of 2 ¥ 10-3) when

Journal compilation © 2008 Blackwell Publishing Ltd, Molecular Microbiology, 67, 719–728 No claim to original US government works

724 C. L. Bair et al. 䊏

Table 1. lcD-CUE and lkD-Ubiquitin association in the presence of a phage display library. lkD-CUE titer

lkD-Ubiquitin titer

lD-library titera

CmlR/KanR lysogensb

106 106 106 106 106 106 106 106

100 250 2500 5000 100 250 2500 5000

0 0 0 0 106 106 106 106

57 108 337 422 53 97 284 404

a. lqprostate cancer fusion display phage library. b. An average of five experiments. In order to assay for potential interference, dual infection and lysogen formation by a presumed ‘bait’ phage and a presumed ‘prey’ phage was measured by adding increasing numbers of bait phage (100– 5000) plus a constant member of the prey phage (106) in the absence and presence of an excess of a phage display library (106). As shown, there was no apparent interference by the presence of the 106 display phage library on lkD-CUE:D-lcUbiquitin associations, as assessed by CmlR/KanR lysogens.

challenged with 200-fold excess of vector phages (data not shown). Similarly, when challenged with a large number of library display phages lacking cml R or kan R antibiotic resistance markers, lcD-CUE and lkD-Ubiquitin were still able to ‘find’ each other. In this assay, lcD-CUE was kept at a constant input (1 ¥ 106) while lkD-Ubiquitin was added in increasing numbers (from 100 to 5000) in the presence or absence of 106 non-antibiotic library display phage. As shown, there appeared to be no interference due to the presence of vast excess of library phage partners on either lkCUE:lcUbiquitin associations. The addition of 50 mg ml-1 BSA or 20% DMSO also did not have any impact on protein–protein association between lcD-CUE:lkD-Ubiquitin (data not shown).

biotic resistant lysogens are shown in Table 2. The prophages were rescued from such lysogens following induction as described in Experimental procedures, purified as lk resistant phages by single plaque isolations and sequenced to reveal the identity of the displayed ‘prey’ protein participating in the interacting processes. We obtained a total of eight double antibiotic resistant lysogens containing potential prey prophages (two by the lcD-CUE and six by the lcD-Ubiquitin phages). Among those captured by lcD-Ubiquitin, the sequences represent segments of two ORFs of unknown functions, NIH_MGC_72 and cDNA clone TKIDN2010115, as well as three known proteins, amyloid beta A4 precursor protein (obtained twice), general transcription factor IIIC polypeptide 2 and EH domain binding protein 1-like-1. Those captured by lcD-CUE represent segments of two ORFs, one of known function (nischarin) and one of unknown function (cDNA clone HE0670). The amyloid beta A4 precursor protein and the EH domain in binding protein 1-like-1 are known to be involved in trafficking during protein destruction, potentially interacting with Ubiquitin in the process. In fact, the DHBP1L1 protein has been shown to bind Ubiquitin through coiled-coil domains (Prag et al., 2005; Stamenova et al., 2007). Nischarin, an imidazoline receptor, is also known to interact with its binding partners through coiled-coiled domains (Lim and Hong, 2004). Given that the gene representation in our library may be limited, it is encouraging that the 2l method fished out proteins that hold functional correlations to the CUE and Ubiquitin proteins, and are demonstrable Ubiquitin binding partners. In order to verify that the interactions discovered are not false positives, dilutions studies were conducted using the potential interacting partner,

Table 2. Use of lkD-CUE and lcD-Ubiquitin phages to capture targets in a prostate display library.

Library panning for novel binding partners The ability of two interacting display phages to interact with each other even in the presence of excess of nonspecific phages prompted us to individually use the two display phages described here (lcD-CUE and lcDUbiquitin) to individually pan against a library of lk display phages constructed from a prostate cancer cDNA library as in Experimental procedures. We note that some of the eukaryotic proteins may not be folded properly within E. coli, and that the representation of a given gene is unknown. We tested the 2l system for the ability of a ‘bait’ display phage to find specific binding partners within the library. The bait phage, lcD-CUE or lcD-Ubiquitin was kept at a constant input of 106, and the lkD-library phages were added at increasing numbers from 101 to 104 as measured by pfu. The lkD-library phages that interacted with lcD-CUE or lcD-Ubiquitin resulting in double anti-

Bait phage titer

lD-library titera

# CmlR/KanR lysogens

lcD-CUE 106 106 106 106

10 100 1 000 10 000

0 0 0 2

lcD-Ubiquitin 106 106 106 106

10 100 1 000 10 000

0 0 1 5

a. lk prostrate cancer fusion display phage library. Aliquots of lkD-CUE and lcD-Ubiquitin display phages were separately incubated with a given number of display phages in the library, followed by addition of bacterial cells as described in Experimental procedures. The mixtures were plated on chloramphenicol/ kanamycin-supplemented LB plates. Shown are the numbers of CmlR/KanR lysogens formed in the two cases.

Journal compilation © 2008 Blackwell Publishing Ltd, Molecular Microbiology, 67, 719–728 No claim to original US government works

A phage display system 725

Fig. 5. The formation CmlR/KanR lysogens at different moi of premixed lcD-DUE and lcD-Ubiquitin display phages with the prey library display phages lkD-nischarin (designated as lkD-L6 above) and lkD-EHBP1L1 (designated at lkD-L8 above). Aliquots of phage preparations at each dilution were mixed, each mixture used to infect 108 host cells, and then plated on double antibiotic plates as described in Experimental procedures. The addition of 1 ¥ 10x moi of phage #1 with 1 ¥ 10x moi of phage #2 result in a total moi of 2 ¥ 10x. The graphs represent the data in log scale. The moi of 2 and 3 on the x-axis are not in scale with those below a moi of 1.

lcD-Ubiquitin and lkD-EHBP1L1 (representing the C-terminal end of the protein), and the pair, lcD-CUE and lkD-nischarin (also representing the C-terminal end of the protein). As shown in Fig. 5, there were significant interactions at low moi between the lkD-Ubiquitin and lkDEHBP1L1, and between lcD-CUE and lkD-nischarin. There were no interactions at low moi between lkDUbiquitin and lkD-EHBP1L1. But there were weak interactions observed between Ubiquitin and nischarin at moi nearing 0.001. The latter interactions may be due to the native ability of Ubiquitin to bind a number of proteins. These studies further verify the specificity as well as the validity of the interactions found using our display phage as bait to pan against a large number of proteins. The clone in lkD-EHBP1L1 is designated as lkD-L8 and that in lkD-nischarin as lkD-L6.

Discussion Phage lambda has been used previously as a display vector showing certain advantages over M13 and T7 phage-based display (Mikawa et al., 1996; Terry et al., 1997; Hoess, 2001; Ansuini et al., 2002; Konthur and Crameri, 2003; Schoonbroodt et al., 2005). We report here the development of a simple 2-Hybrid system based on phage l D-display that takes advantage of l biology (Kourilsky, 1973; Kourilsky and Knapp, 1974) and the use of antibiotic selection. We surmised that if two or more phages are given a reason to associate (here, through their displayed peptides) they will coinfect a cell at low moi (10-2-10-6) and produce lysogens. We provided a convenient selection for lysogenic cells by introducing a different antibiotic resistance marker, cml R or kan R, in each of the two display phages. This 2l system therefore represents an in vitro platform that holds many advantages

over other cellular, phage and immobilization systems (Chowdhury and Pastan, 1999; Halperin et al., 2003; Suzuki et al., 2005; Yu et al., 2005). Using this technique, we successfully demonstrated specific associations between known or anticipated binding partners. The specificity and sensitivity observed allowed us to perform library panning wherein several potentially specific clones were fished out from a library using CUE and Ubiquitin as bait, showing interactions between functionally correlated or previously demonstrated interacting partners. The 2l system also demonstrated the interactions of larger, heterodimeric complexes using the highly structured eukaryotic proteins, S100b, p53 and HDM2 (a human homologue of MDM2) (Wilder et al., 2007). The S100b– p53 interaction was successfully screened for inhibitory drugs using a panel of small molecules to confirm studies with free proteins in vitro (Wilder et al., 2007). In contrast to established l display techniques for biopanning using immobilized proteins, our approach does not require extensive and elaborate protein immobilization, or harsh chemical treatments that can prevent or disrupt protein binding and potentially destroy target peptides. Limitations such as poor folding in vivo (i.e. disulphide bonds that require a reducing environment) can be overcome by providing all necessary conditions in solution. The associations can be performed in a wide range of conditions without affecting the infection process, and every aspect of the reaction (time, temperature, pH, salt concentrations, inhibitors, mediator-factors, other small molecules, etc.) can be controlled to a high degree. We anticipate that the 2l system holds potential for applications in (i) assaying known or suspected binding partners, especially those that are recalcitrant to other display platforms (Malik et al., 1996; Santini et al., 1998; Yang et al., 2000; Lunder et al., 2005), (ii) simple and

Journal compilation © 2008 Blackwell Publishing Ltd, Molecular Microbiology, 67, 719–728 No claim to original US government works

726 C. L. Bair et al. 䊏

economic screening for inhibitors of a binding pair (drug screening), (iii) improving the affinity between two binding partners further by mutagenizing one or both of the partners and looking for double antibiotic resistant lysogens at lower moi, (iv) more easily deciphering the various binding partners within a protein multicomplex (Krauss et al., 2003; Ho et al., 2005), (v) optimizing in vitro binding conditions through the addition of ions, mediators, chaperones, etc. While the 2l system does possess intrinsic advantages, it may have limitations in quantification. It may be restricted to studying the changes in affinity due to alteration of one partner within a given pair rather than comparing the strengths between pairs due to such variables as differing levels of multivalency, steric hindrances, etc. We suspect that display of proteases that can digest the outer lambdoid proteins, proteins that require specific modifications for proper activity (i.e. glycosylation, phosphorylation, and other eukaryotic post-translation modifications), very large proteins whose gene size prohibit cloning with the vector phage, antibacterial proteins, etc. may not be workable in our system. However, as the D-protein can be added to D-less virions (Zanghi et al., 2005), some of these obstacles can be overcome by incorporating separately prepared D-fusion proteins into phage in vitro.

Experimental procedures Plasmids Plasmid pVCDcDL3 (Fig. 2) has been described in Gupta et al. (2003). It contains the selectable amp R gene and the l phage virion major capsid protein, D, fused to a lac promoter and gene. A multiple cloning site is present at the carboxy terminus of the D gene for the purpose of cloning fusion display peptides or proteins. This construct generates display peptides fused at the amino terminus of the virion capsid protein, allowing the amino terminal of the fusion peptide to remain open for binding activities. It is known that proteins can be displayed on phage l virion through either unstructured termini of the D protein (Malik et al., 1996).

Bacteriophage strains Bacteriophage lDL1 was described by Gupta et al. (2003). It was derived from lDam imm21 nin5 Dam4 (Kay et al., 1996). The lc and lk are derivatives of lDL1, and were constructed during this study by l-Ked mediated recombination (Court et al., 2002) a cml R and kan R cassette, respectively, between the phage R gene and the right cos end (see Fig. 2).

Bacterial strains Escherichia coli LE392 (supE+,F+), E. coli W3110 (sup -), E. coli BM25.8 (supE+, lox-Cre+) and E. coli DY330 [W3110 DlacU169 gal490 pglD8 cI857 D(cro-bioA)].

Microbiological methods All microbiological methods as well as phage biological methods follow the protocols described in Silhavy et al. (1984), Court et al. (2002) and Gupta et al. (2003).

Proteins and peptides Vps9-CUE domain of wild type and mutant proteins were expressed and purified from E. coli as described previously (Prag et al., 2003). Bovine Ubiquitin was purchased from Sigma. The 17-amino-acid residue acidic aptamer (Glu-Gly)4 and the basic aptamer (Gly-Arg)4 were purchased from New England Peptide (Ipswitch, MA). These contain the leader sequence GSGPVGPGG followed by the eight amino acidic or basic aptamer.

Construction of l D-fusion plasmids The PCR amplified DNA sequences of Acid, Base, CUE and Ubiquitin genes used in this study were cloned into the mcs of plasmid vector pDC3 so that the ORFs of l D and the fusion polypeptides are in the same frame. The successful D-fusions in the plasmids were then transferred to phage l vectors, lc or lk, by Cre-lox mediated recombination (Gupta et al., 2003). The l D-fusion phages (i.e. display phages) were isolated by selecting for AmpR l lysogens that are Lac- in a lacZa- host E. coli TG1 (Gupta et al., 2003). The lc and lk vector phages are lacZa+ because they contain the lacZa gene at the site into which the D-fusion from the plasmid is recombined. The gene contents of the display phages were confirmed by DNA sequencing of the fused genes in the phages. Construction and preparation of l D-display phages: overnight cultures of E. coli BM25.8/pDC3-X were diluted in Luria–Bertani (LB) supplemented with 30 mg ml-1 Amp and 12.5 mg ml-1 Cml and grown to mid-log phase. The cells were collected at 4400 g for 7 min then resuspended in 1 ml of diluted samples of lK or lC vector phage at a final count of 1 ¥ 108 pfu ml-1 in TMG buffer (Tris·HCl; pH 7.0; MgSO4·7H2O; Gelatin). Infection was allowed to occur for 5 min at 25°C, followed by 4 h at 37°C to allow recombination between the gene of interest and the lk or lc vector phage (Fig. 2). The details of the genetics in this process are covered in Gupta et al. (2003). Finally lysis was induced by temperature shifting to 42°C. The cleared lysate was filtered through a 0.22 mm membrane (Millipore). Approximately 150 ml of this lysate was used to infect 5 ml of E. coli TG1 (supE +) grown in LB; 0.4% maltose; 10 mM MgCl2 to an OD600 = 0.5. The class of wild type D was verified by the poor plaque forming ability on E. coli W3110 (sup-). After 15 min at 37°C, the cells were spread on agar plates supplemented with either Cml/Amp or Kan/Amp antibiotics, and incubated overnight at 32°C. Single colonies were isolated, re-streaked on the corresponding double antibiotic plates and incubated overnight at 32°C. Colonies were then cultured in LB or LB supplemented with 30 mg ml-1 ampicillin overnight at 32°C. The following day, the cultures were diluted 100 times in LB, and grown for 4 h at 32°C. Phages were then induced to lyse by temperature shifting to 42°C, and the cleared lysates were filtered through 0.22 mm membranes (Millipore). Individual

Journal compilation © 2008 Blackwell Publishing Ltd, Molecular Microbiology, 67, 719–728 No claim to original US government works

A phage display system 727

phage were purified by two or more rounds of single plaque isolation at 42°C, and lysates made. The purified lysates were assayed by spot tests on E. coli TG1 on MacConkeyAMP agar; 50 mg ml-1 ampicillin; 1% lactose plates for ampicillin resistance and the loss of the LacZa gene. The desired recombinant phages were verified by DNA sequencing.

2l assays Pairs of l display phages were assayed for their ability to specifically associate with each other. First, the phages were diluted individually by logs in association buffer (20 mM Tris·HCl; pH 7.4; 10 mM CaCl2; 10 mM MgCl2; 100 mM NaCl) to yield a series of moi to 108 cells. After 15 min at 37°C, 200 ml of log-phase E. coli TG1 was added, further incubated for 15 min at 37°C, and then the entire reaction was spread on LB-agar supplemented with 12.5 mg ml-1 Cml and/or 25 mg ml-1 Kan. Plates were incubated overnight at 32°C for lysogen formation.

Acknowledgements This research was supported by the Intramural Research Program of the National Institutes of Health, National Cancer Institute, Center for Cancer Research, and the National Institute of Allergy and Infectious Diseases Biodefense Intramural Research Fund. The authors would like to thank Don Court, NCI, for his guidance and advice during the development of the 2l system, and Dom Esposito of FCRF-NCI, Frederick, MD for constructing a pDC3 plasmid derivative that is capable of recombination with the Gateway library vectors for creation of the prostate cancer lambda fusion display phage library.

References Ansuini, H., Cicchini, C., Nicosia, A., Tripodi, M., Cortese, R., and Luzzago, A. (2002) Biotin-tagged cDNA expression libraries displayed on lambda phage: a new tool for the selection of natural protein ligands. Nucleic Acids Res 30: e78. Auerbach, D., Thaminy, S., Hottiger, M.O., and Stagljar, I. (2002) The post-genomic era of interactive proteomics: facts and perspectives. Proteomics 2: 611–623. Chowdhury, P.S., and Pastan, I. (1999) Improving antibody affinity by mimicking somatic hypermutation in vitro. Nat Biotechnol 17: 568–572. Cicchini, C., Ansuini, H., Amicone, L., Alonzi, T., Nicosia, A., Cortese, R., et al. (2002) Searching for DNA–protein interactions by lambda phage display. J Mol Biol 322: 697–706. Court, D.L., Sawitzke, J.A., and Thomason, L.C. (2002) Genetic engineering using homologous recombination. Annu Rev Genet 36: 361–388. Dove, S.L., and Hochschild, A. (2004) A bacterial two-hybrid system based on transcription activation. Methods Mol Biol 261: 231–246. Fields, S., and Song, O. (1989) A novel genetic system to detect protein–protein interactions. Nature 340: 245–246. Gupta, A., Onda, M., Pastan, I., Adhya, S., and Chaudhary,

V.K. (2003) High-density functional display of proteins on bacteriophage lambda. J Mol Biol 334: 241–254. Hagen, E.S., Gray, C.L., and Kuijper, J.L. (1988) Assaying the quality of cDNA libraries. Biotechniques 6: 340–345. Halperin, I., Wolfson, H., and Nussinov, R. (2003) SiteLight: binding-site prediction using phage display libraries. Protein Sci 12: 1344–1359. Ho, M., Kreitman, R.J., Onda, M., and Pastan, I. (2005) In vitro antibody evolution targeting germline hot spots to increase activity of an anti-CD22 immunotoxin. J Biol Chem 280: 607–617. Hoess, R.H. (2001) Protein design and phage display. Chem Rev 101: 3208–3218. Hoess, R.H. (2002) Bacteriophage lambda as a vehicle for peptide and protein display. Curr Pharm Biotechnol 3: 23–28. Joung, J.K., Ramm, E.I., and Pabo, C.O. (2000) A bacterial two-hybrid selection system for studying protein-DNA and protein–protein interactions. Proc Natl Acad Sci USA 97: 7382–7387. Karimova, G., Pidoux, J., Ullmann, A., and Ladant, D. (1998) A bacterial two-hybrid system based on a reconstituted signal transduction pathway. Proc Natl Acad Sci USA 95: 5752–5756. Kay, B.K., Winter, J., and McCaffrey, J. (1996) Phage Display of Peptides and Proteins. New York: Academic Press. Kirsch, M., Zaman, M., Meier, D., Dubel, S., and Hust, M. (2005) Parameters affecting the display of antibodies on phage. J Immunol Methods 301: 173–185. Konthur, Z., and Crameri, R. (2003) High-throughput applications of phage display in proteomic analysis. Targets 2: 261–270. Kourilsky, P. (1973) Lysogenization by bacteriophage lambda. I. Multiple infection and lysogenic response. Mol Gen Genet 122: 183–195. Kourilsky, P., and Knapp, A. (1974) Lysogenization by bacteriophage lambda. III. Multiplicity dependent phenomena occurring upon infection by lambda. Biochimie 56: 1517– 1523. Krauss, J., Arndt, M.E.A., Martin, A.C.R., Liu, H., and Rybak, S.M. (2003) Specificity grafting of human antibody frameworks selected from a phage display library: generation of a highly stable humanized anti-CD22 single-chain Fv fragment. Protein Eng 16: 753–759. Li, J.J., and Herskowitz, I. (1993) Isolation of ORC6, a component of the yeast origin recognition complex by a onehybrid system. Science 262: 1870–1874. Lim, P.K., and Hong, W. (2004) Nischarin/imidazoline receptor antisera-selected protein is targeted to the endosomes by a combined action of a PX domain and a coiled-coil region. J Biol Chem 279: 54770–54782. Lunder, M., Bratkovic, T., Doljak, B., Kreft, S., Urieb, U., Strukelj, B., et al. (2005) Comparison of bacterial and phage display peptide libraries in search of target-binding motif. Appl Biochem Biotechnol 127: 125–132. Malik, P., Terry, T.D., Gowda, L.R., Langara, A., Petukhov, S.A., Symmons, M.F., et al. (1996) Role of capsid structure and membrane protein processing in determining the size and copy number of peptides displayed on the major coat protein of filamentous bacteriophage. J Mol Biol 260: 9–21. Mikawa, Y.G., Maruyama, I.N., and Brenner, S. (1996)

Journal compilation © 2008 Blackwell Publishing Ltd, Molecular Microbiology, 67, 719–728 No claim to original US government works

728 C. L. Bair et al. 䊏

Surface display of proteins on bacteriophage lambda heads. J Mol Biol 262: 21–30. Oppenheim, A., Kobiler, O., Stavans, J., Court, D., and Adhya, S. (2005) Switches in bacteriophage lambda development. Annu Rev Genet 39: 409–429. Prag, G., Misra, S., Jones, E.A., Ghirlando, R., Davies, B.A., Horazdovsky, B.F., et al. (2003) Mechanism of Ubiquitin recognition by the CUE domain of Vps9p. Cell 113: 609– 620. Prag, G., Lee, S., Mattera, R., Arighi, C.N., Beach, B.M., Bonifacino, J.S., et al. (2005) Structural mechanism for ubiquitinated-cargo recognition by the Golgi-localized, gamma-ear-containing, ADP-ribosylation-factor-binding proteins. Proc Natl Acad Sci USA 102: 2334–2339. Santini, C., Brennan, D., Mennuni, C., Hoess, R.H., Nicosia, A., Cortese, R., et al. (1998) Efficient display of an HCV cDNA expression library as C-terminal fusion to the capsid protein D of bacteriophage lambda. J Mol Biol 282: 125– 135. Schoonbroodt, S., Frans, N., DeSouza, M., Eren, R., Priel, S., Brosh, N., et al. (2005) Oligonucleotide-assisted cleavage and ligation: a novel directional DNA cloning technology to capture cDNAs. Application in the construction of a human immune antibody phage-display library. Nucleic Acids Res 33: e81. Shih, S.C., Prag, G., Francis, S.A., Sutanto, M.A., Hurley, J.H., and Hicke, L. (2003) A Ubiquitin binding motif required for intramolecular monoubiquitylation, the CUE domain. EMBO J 22: 1273–1281. Silhavy, T.J., Berman, M.L., and Fritsch, E.F. (1984) Experiments with Gene Fusions. Cold Spring Harbor, NY: Cold Spring Harbor Press. Stamenova, S.D., French, M.E., He, Y., Francis, S.A., Kramer, Z.B., and Hicke, L. (2007) Ubiquitin binds to and regulates a subset of SH3 domains. Mol Cell 25: 273–284. Sternberg, N., and Hoess, R.H. (1995) Display of peptides and proteins on the surface of bacteriophage lambda. Proc Natl Acad Sci USA 92: 1609–1613.

Suzuki, Y., Ito, S., Otsuka, K., Iwasawa, E., Nakajima, M., and Yamaguchi, I. (2005) Preparation of functional singlechain antibodies against bioactive gibberellins by utilizing randomly mutagenized phage-display libraries. Biosci Biotechnol Biochem 69: 610–619. Terry, T.D., Malik, P., and Perham, R.N. (1997) Accessibility of peptides displayed on filamentous bacteriophage virions: susceptibility to proteinases. Biol Chem 378: 523– 530. Wilder, P.T., Lin, J., Bair, C.L., Charpentier, T.H., Yang, D., Liriano, M., et al. (2007) Recognition of the tumor suppressor protein p53 and other protein targets by the calciumbinding protein S100B. Biochim Biophys Acta 1763: 1284– 1297. Yang, F., Forrer, P., Dauter, Z., Conway, J.F., Cheng, N., Cerritelli, M.E., et al. (2000) Novel fold and capsid-binding properties of the lambda-phage display platform protein gpD. Nat Struct Biol 7: 230–237. Yu, B., Ni, M., Li, W.H., Lei, P., Xing, W., Xiao, D.W., et al. (2005) Human scFv antibody fragments specific for hepatocellular carcinoma selected from a phage display library. World J Gastroenterol 11: 3985–3989. Zanghi, C.N., Lankes, H.A., Tretheway, B.B., Wegman, J., and Dewhurst, S. (2005) A simple method for displaying recalcitrant proteins on the surface of bacteriophage lambda. Nucleic Acids Res 33: e160. Zhang, J., and Lautar, S. (1996) A yeast three-hybrid method to clone ternary protein complex components. Anal Biochem 242: 68–72. Zhu, W., Williams, R.S., and Kodadek, T. (2000) A CDC6 protein-binding peptide selected using a bacterial twohybrid-like system is a cell cycle inhibitor. J Biol Chem 275: 32098–32105. Zucconi, A., Dente, L., Santonico, E., Castagnoli, L., and Cesareni, G. (2001) Selection of ligands by panning of domain libraries displayed on phage lambda reveals new potential partners of synaptojanin. J Mol Biol 307: 1329– 1339.

Journal compilation © 2008 Blackwell Publishing Ltd, Molecular Microbiology, 67, 719–728 No claim to original US government works