A β-Galactosidase-Based Bacterial Two-Hybrid System To Assess

A β-Galactosidase-Based Bacterial Two-Hybrid System To Assess Protein-Protein Interactions in the Correct Cellular Environment Jimmy Borloo,* Lina De Smet, Bjorn Vergauwen, Jozef J. Van Beeumen, and Bart Devreese Department of Biochemistry, Physiology and Microbiology. Ghent University. Laboratory for Protein Biochemistry and Protein Engineering. K.L. Ledeganckstraat 35, 9000 Ghent, Belgium Received January 23, 2007

The vast majority of proteins functions in complex with one or more of the same or other proteins, indicating that protein-protein interactions play crucial roles in biology. Here, we present a β-galactosidase reconstitution-based bacterial two-hybrid system in which two proteins of interest are fused to two non-functional but complementing β-galactosidase truncations (∆R and ∆ω). The level of complemented β-galactosidase activity, driven by the protein-protein recognition between both nonβ-galactosidase parts of the chimeras, reflects whether or not the proteins of interest interact. Our approach was validated by reconfirming some well-established Escherichia coli cytoplasmic and membranous interactions, including well-chosen mutants, and providing the first in vivo evidence for the transient periplasmic interaction between Rhodobacter capsulatus cytochrome c2 and cytochrome c peroxidase. We demonstrated the major advantages of this in vivo two-hybrid technique: i) analyses of interactions are not limited to particular cellular compartments, ii) the potential of using the system in mutation-driven structure-function studies, and iii) the possibility of its application to transiently interacting proteins. These benefits demonstrate the relevance of the method as a powerful new tool in the broad spectrum of interaction assessment methods. Keywords: protein-protein interactions • bacterial two-hybrid system • β-galactosidase

Introduction The elucidation of protein-protein interactions is subject to intense research and currently constitutes one of the highest priorities in the attempt to understand some particular biological processes. Many techniques have been developed to identify these interactions; among them are in vitro techniques such as pull-down analyses and co-immunoprecipitation, as well as in vivo strategies such as fluorescence resonance energy transfer (FRET).1 These techniques have proven to be successful and offer major advantages such as sensitivity in the case of FRET and a pragmatic character when considering pull-down analyses. These techniques also bear serious drawbacks; the characteristic high sensitivity of FRET, for instance, results in significant background signals, whereas co-immunoprecipitation and pull-down analyses require specific antibodies and harsh elution conditions, respectively. A popular technique, developed in the late 1980s and early 1990s and nowadays often used in so-called ‘functional genomics’, is the yeast two-hybrid system.2,3 This technique is based on the coexpression of two chimeras or hybrid proteins, each consisting of a protein of interest and either a DNA binding domain or a transcription activation domain. When the two proteins of interest associate with each other, a fully * Corresponding author. Laboratory for Protein Biochemistry and Protein Engineering, Ghent University, K.L. Ledeganckstraat 35, B-9000 Ghent, Belgium. Phone: +32 9 264 51 26. Fax: +32 9 264 53 38. E-mail: Jimmy. [email protected]. 10.1021/pr070037j CCC: $37.00

 2007 American Chemical Society

functional transcription factor is formed and a reporter gene is transcribed, leading to an easily detectable phenotype. Since its development, the technique itself has been extensively improved and diversified, giving rise to one- and three-hybrid systems4,5 as well as to reverse n-hybrid systems.6 The yeast two-hybrid system paved the road for functional genomics, that is, the screening of an entire library of proteins (preys), encoded by a given genome, that may form complexes with a particular protein (bait).7,8 Notwithstanding its genius, the yeast twohybrid system has the restriction that all interactions should occur in the nucleus, which limits its use to cytosolic proteins. An alternative method for detecting protein-protein interactions is the use of enzyme complementation, rather than transcriptional activation. Different systems, including β-lactamase, β-galactosidase, and mouse dihydrofolate reductase complementation have been described for use in eukaryotic cells.9-11 Bacterial two-hybrid alternatives have been developed as well. These systems are mostly based on the repression or activation of a reporter gene.12,13 Although a majority of the bacterial two-hybrid systems are still restricted to analyses of cytoplasmic interactions, some have been developed that allow the analysis of periplasmic and membranous protein-protein interactions, such as the Vibrio cholerae cytoplasmic membranelocalized transcriptional regulator ToxR,14 the Bordetella pertussis adenylate cyclase reconstitution system,15 and, analogous to the eukaryotic variant, the β-lactamase reconstitution based Journal of Proteome Research 2007, 6, 2587-2595

2587

Published on Web 06/01/2007

research articles

Borloo et al.

Table 1. Plasmids Used in This Study plasmid

pCR2.1-TOPO pCR-XL-TOPO pALTER-Ex1 pALTER-Ex1∆BamHI pACYCDuet-1 pETDuet-1 pETDuet-1∆SphI pACYCDuet-1Ωtac pETDuet-1∆SphIΩtac pB2H∆R pB2H∆ω pB2H∆RΩamyA pB2H∆RΩdnaN pB2H∆RΩfimC pB2H∆RΩc2 pB2H∆RΩatpE pB2H∆RΩatpF pB2H∆ωΩdnaE pB2H∆ωΩfecB pB2H∆ωΩCCP pB2H∆ωΩfhuA pB2H∆ωΩatpB

description

3.9-kb vector for cloning PCR products; 3.5-kb vector for cloning long PCR products (3-10 kb); KmR 5.8-kb cloning vector; used as source of the tac promoter; TetR pALTER-Ex1 with the BamHI restriction site deleted; 5.8-kb; TetR 4.0-kb vector designed for coexpression of two target genes; CmR 5.4-kb vector designed for coexpression of two target genes; CbR pETDuet-1 with the SphI restriction site deleted; 5.4-kb; CbR pACYCDuet-1 with the tac promoter from pALTER-Ex1∆BamHI cloned in the EcoRV-PstI site; 4.5-kb; CmR pETDuet-1∆SphI with the tac promoter from pALTER-Ex1∆BamHI cloned in the EcoRV-PstI site; 6.0-kb; CbR pACYCDuet-1Ωtac with the E. coli β-galactosidase fragment lacking the sequence for amino acids 11-41 (∆a) cloned in the BamHI-NcoI site; CmR; 7.5-kb pETDuet-1∆SphIΩtac with the E. coli β-galactosidase fragment lacking the sequence for amino acids 789-1023 (∆ω) cloned in the BamHI-NcoI site; CbR; 8.3-kb pB2H∆R with the amyA gene cloned in the BamHI-SphI site; CmR; 9.0-kb pB2H∆R with the dnaN gene cloned in the BamHI-SphI site; CmR; 8.6-kb pB2H∆R with the fimC gene cloned in the BamHI-SphI site; CmR; 8.2-kb pB2H∆R with the cytochrome c2 gene cloned in the BglII-SphI site; CmR; 7.9-kb pB2H∆R with the atpE gene cloned in the BamHI-SphI site; CmR; 7.7-kb pB2H∆R with the atpF gene cloned in the BamHI-SphI site; CmR; 8.0-kb pB2H∆ω with the dnaE gene cloned in the BamHI-SphI site; CbR; 11.8-kb pB2H∆ω with the fecB gene cloned in the BamHI-SphI site; CbR; 9.2-kb pB2H∆ω with the CCP gene cloned in the BamHI-SphI site; CbR; 9.3-kb pB2H∆ω with the fhuA gene cloned in the BamHI-SphI site; CbR; 10.5-kb pB2H∆ω with the atpB gene cloned in the BglII-SphI site; CbR; 9.1-kb

two-hybrid system.16 Here, we present a bacterial two-hybrid system which is based on trans complementation of β-galactosidase, the product of the lacZ gene that is responsible for the breakdown of lactose and other β-galactosides into monosaccharides. The biologically active β-galactosidase enzyme is a tetramer of four 116-kDa monomers (A-D), each encompassing 1023 amino acids and composed of five domains (1-5) of which domain 3 has a ‘TIM’ barrel structure with the active site at its C-terminal end. A major part of this active site is formed by a deep cleft in the core of the barrel. The active site is further constituted of loops from the first and fifth domain of the same monomer as well as of a loop from domain 2 of a neighboring monomer. Therefore, individual monomers cannot catalyze substrate hydrolysis. Studies conducted almost four decades ago revealed that both the R (N-terminal part) and ω (C-terminal part) fragments of the enzyme are necessary for the formation and assembly of an active β-galactosidase and that mutants lacking one or the other can successfully complement each other.17,18 This feature has found its applications in biotechnology, for example, blue-white screening in cloning experiments and the complementation assay for protein translocation (CAPT).19 We here describe the design and validation of an Escherichia coli β-galactosidase reconstitution-based twohybrid system. By testing the method with properly selected protein couples from each cell compartment of E. coli and Rhodobacter capsulatus, we show that the technique is amenable to protein-protein interactions occurring in the cytoplasm, the periplasm, and the membrane of a bacterial cell.

Experimental Section Bacterial Strains, Plasmids, Media. and Growth Conditions. E. coli strain XL-1 Blue (Stratagene, La Jolla, CA) was used in all subcloning steps, whereas recombination-deficient strain JM109 (The Coli Genetic Stock Center, New Haven, CT) was applied in all subsequent cloning experiments. β-Galactosidase activity tests were performed using E. coli strain MC1061 (The Coli Genetic Stock Center), which lacks the entire lacZ locus. 2588

KmR

Journal of Proteome Research • Vol. 6, No. 7, 2007

source

Invitrogen Invitrogen Promega This work Novagen This work This work This work This work This work This work This work This work This work This work This work This work This work This work This work This work This work

All plasmids used and constructed during this study are summarized in Table 1. E. coli cultures were grown aerobically in Luria Bertani broth20 on a rotary shaker (200 rpm) or anaerobically in Minimal Salt Medium, both at 37 °C.21 Growth media were supplemented with appropriate antibiotics when necessary, including chloramphenicol (Cm) at 25 µg/mL, carbenicillin (Cb) at 100 µg/mL, and kanamycin (Km) at 25 µg/ mL. When required, IPTG (Duchefa, Haarlem, The Netherlands) was added to a final concentration of 20 mM. DNA Manipulations. A list of the synthetic oligonucleotides used in this study is presented in Table 2. Oligonucleotides included the restriction sites used to clone the DNA fragments into the vectors. Restriction digests, cloning, and DNA electrophoresis were performed using standard techniques.20 DNA ligations were performed using T4 DNA Ligase (Promega, Madison, WI). Restriction sites were eliminated using T4 DNA Polymerase or Mung Bean Nuclease (both from New England Biolabs, Ipswich, MA) according to the supplier’s recommendations. Isolation of plasmid DNA was accomplished using the QIAprep Plasmid Midi Kit 100 (Qiagen, Hilden, Germany). All DNA constructs were confirmed by DNA sequencing (GENOME Express, Meylan, France). Construction of the Vectors. The compatible pACYCDuet-1 and pETDuet-1 vectors are central in our bacterial two-hybrid system. Since most of the test proteins in the development of this system were cloned into the BamHI-SphI restriction sites of the pB2H∆R and pB2H∆ω constructs (for exceptions, see Table 2), all unsuitable endogenous BamHI and SphI restriction sites were removed from the starting vectors. The sequences coding for the test proteins AmyA, DnaE, DnaN, FecB, FimC, cytochrome c2, cytochrome c peroxidase (CCP), FhuA, AtpB, AtpE, and AtpF were obtained through PCR using genomic DNA as template and the primers listed in Table 2. The PCR products were subcloned into pCR2.1-TOPO or pCR-XL-TOPO (Invitrogen, Carlsbad, CA) and subsequently cloned into pB2H∆R and pB2H∆ω. The genes amyA, dnaN, fimC, cytochrome c2, atpE, and atpF were cloned in frame with ∆R in pB2H∆R, while dnaE,

research articles

β-Galactosidase-Based Bacterial Two-Hybrid System Table 2. Test Proteins and Synthetic Oligonucleotides Used in This Study protein

∆R ∆ω AmyA DnaE DnaN FecB FimC c2 CCP FhuA AtpF AtpB AtpE DnaE917 DnaE1153 DnaE920/1153

oligonucleotide name and sequencea

description

E. coli XL-1 Blue β-galactosidase lacking amino acids 11-41 (cytoplasmic) E. coli β-galactosidase lacking amino acids 789-1023 (cytoplasmic) E. coli R-amylase (cytoplasmic) E. coli DNA polymerase III, β-subunit (cytoplasmic) E. coli DNA polymerase III, β-subunit (cytoplasmic) E. coli citrate-dependent iron transport protein (periplasmic) E. coli chaperone, required for type I fimbrae (periplasmic) Rhodobacter capsulatus mono-heme cytochrome c2 (periplasmic) R. capsulatus di-heme cytochrome c peroxidase (periplasmic) E. coli protein receptor for ferrichrome, colicin M and phages T1, T5, and phi80 (outer membrane) E. coli ATP-synthase, F0 sector, subunit B (inner membrane) E. coli ATP synthase, F0 sector, subunit A (inner membrane) E. coli ATP-synthase, F0 sector, subunit C (inner membrane) DnaE mutant lacking the entire C-terminal part starting from amino acid 918 to 1159 DnaE mutant lacking the C-terminal part consisting of amino acids 1154-1159 DnaE mutant lacking lacking amino acids 920-924 and 1154-1159

A1 A2 B1 B2 C1 C2 D1 D2 E1 E2 F1 F2 G1 G2 H1 H2 I1 I2 J1 J2

5′-TCTAGAGGATCCGGGGGTATGACCATGATTACGGATTCA-3′ 5′-CCATGGTTATTTTTGACACCAGACCAAC-3′ 5′-TCTAGAGGATCCGGGGGTATGACCATGATTACGGATTCA-3′ 5′-CCATGGTTATGCACGGGTGAACTGATCGCG-3′ 5′-GCATGCAGGAGGACAGCTATGCGTAATCCCACGCTGTTACAATGT-3′ 5′-GGATCCAATCACCTCTTCGATAACCCACACGCT-3′ 5′-GCATGCAGGAGGACAGCTATGTCTGAACCACGTTTCGTACACCTG-3′ 5′-GGATCCGTCAAACTCCAGTTCCACCTGCTCCGAA-3′ 5′-GCATGCAGGAGGACAGCTATGAAATTTACCGTAGAACGTGAGCAT-3′ 5′-GGATCCCAGTCTCATTGGCATGACAACATAAGC-3′ 5′-GCATGCAGGAGGACAGCTATGATTATGTTGGCATTTATCCGTTTT-3′ 5′-GGATCCTTTCACAACGGTAAGCGGCTGATGGTG-3′ 5′-GCATGCAGGAGGACAGCTATGAGTAATAAAAACGTCAATGTAAGG-3′ 5′-GGATCCTTCCATTACGCCCGTCATTTTGGGGGT-3′ 5′-GCATGCATGAAGATCAGCCTCA-3′ 5′-AGATCTTTTCACGACCGAGGCC-3 5′-GCATGCATGAAACGCACTCAGA-3′ 5′-GGATCCGTTCATGTGCTCGGGC-3′ 5′-GCATGCAGGAGGACAGCTATGGCGCGTTCCAAAACTGCTCAGCCA-3′ 5′-GGATCCGAAACGGAAGGTTGCGGTTGCAACGAC-3′

K1 K2 L1 L2 M1 M2 N1

5′-GCATGCAGGAGGACAGCTATGAATCTTAACGCAacaatcctcggc-3′ 5′-GGATCCCAGTTCAGCGACAAGTTTATCCAC-3′ 5′-GCATGCAGGAGGACAGCTATGGCTTCAGAAAATATGACGCCGCAG-3′ 5′-AGATCTATGTTCTTCAGACGCCATCGACAGATA-3′ 5′-GCATGCAGGAGGACAGCTATGGAAAACCTGAATATGGATCTGCTG-3′ 5′-GGATCCCGCGACAGCGAACATCACGTACAGACC-3′ 5′-GGATCCAGCTTCCGCTTTCGCGTGTTGATCTG-3′

O1

5′-GGATCCCTCCGAACCAATGAGGCCACGGAGA-3′

P1 P2

5′-GGCGTGCTGGCCGAAGAGCCGGAA-3′ 5′-ACCGATAGCTTCCGCTTTCGCGTG-3′

a Underlined regions indicate the added restriction endonuclease sites to facilitate cloning: BamHI for A1, B1, C2, D2, E2, F2, G2, I2, J2, K2, M2, N1, and O1; SphI for C1, D1, E1, F1, G1, H1, I1, J1, K1, L1, and M1; NcoI for A2 and B2; BglII for H2 and L2.

fecB, CCP, fhuA and atpB were cloned in frame with ∆ω in pB2H∆ω. To test the potential of using our two-hybrid system in mutation-driven structure-function studies and to address fusion protein cleavage issues, deletion mapping analysis was performed using DnaE mutants, lacking the specific amino acids necessary for interaction with DnaN.22,23 DnaE917 and DnaE1153 (for nomenclature, see Table 2) were generated via PCR as mentioned above, whereas a DnaE1153∆920-924 mutant (termed DnaE920/1153) was obtained by applying the pCR-XLTOPO-DnaE1153 construct and the suitable primers, listed in Table 2, in a Phusion site-directed mutagenesis reaction, carried out according to the supplier’s recommendations (Finnzymes, Finland). In a following step, the DnaE1153, DnaE917, and DnaE920/1153 fragments were cloned into pB2H∆ω. Coexpression of the Fusion Proteins. To simultaneously express two fusion proteins within a cell, electrocompetent E. coli MC1061 cells harboring pB2H∆RΩatpF, pB2H∆RΩatpE, pB2H∆ωΩdnaE, pB2H∆ωΩfecB, and pB2H∆ωΩCCP were transformed with the complementary constructs as indicated in Supporting Information Figure 1. Coexpression of the (fusion) proteins was achieved by inoculating the growth medium with the proper double transformant and immediately adding IPTG. The cells were grown overnight before lysis. Expression or coexpression of heme-containing proteins (cytochrome c2 and cytochrome c peroxidase) was obtained by anaerobically growing the cells overnight at 37 °C in Minimal Salt Medium21 immediately supplemented with IPTG. Preparation of Cellular Fractions. Cell lysis was carried out by spinning down the cells for 10 min at 10 000g. The pellet was resuspended in 1 mL water with Complete Protease Inhibitor (Roche Diagnostics Corporation, Indianapolis, IN) (1

tablet per 200 mL of sample), followed by sonication for 30 s using a Branson Digital Sonifier Model 250-D at 10% of its maximum force. When needed, the soluble protein fraction (cytoplasm plus periplasm) was prepared by subsequently removing the insoluble fraction by centrifugation for 5 min at maximum speed in a tabletop centrifuge (Eppendorf, Hamburg, Germany). The supernatant contained the soluble protein fraction of the cells. In due case, the periplasmic fraction was prepared by osmotic shock.24 Briefly, overnight cultures were centrifuged, washed in ice-cold 50 mM potassium phosphate buffer, pH 7.4, resuspended in 1/5 of the original volume in plasmolysis buffer (50 mM Tris, 2.5 mM EDTA, and 20% (w/v) sucrose; pH 7.4), and incubated for 10 min at room temperature. The cells were collected by centrifugation at 20 800g and resuspended in 1/10 of the original volume in ice-cold water. After 10 min of incubation at 4 °C, the cell debris was separated from the periplasmic fraction by centrifugation. The membrane fraction was prepared in the same way as described above for the soluble protein fraction, except that after sonication the insoluble fraction was retained by centrifugation at 20 800g. Subsequently, the pellet was resuspended in 1 mL of water containing 0.5% N-lauroylsarcosine (sarkosyl) (Sigma, St. Louis, MO) and Complete Protease Inhibitor, and was stirred at 37 °C for 1 h. After centrifugation at 20 800g, the supernatant was considered to be the soluble membrane fraction. Enzymatic Assays. β-Galactosidase activity was assayed quantitatively at room temperature by following o-nitrophenylβ,D-galactose (ONPG) hydrolysis and 2-nitrophenol formation at 420 nm in a double beam spectrophotometer (Uvikon, Kontron, Herts, U.K.) in a total volume of 1 mL using β-Galactosidase Assay Buffer (Pierce, Rockford, IL). Half of the Journal of Proteome Research • Vol. 6, No. 7, 2007 2589

research articles volume was protein sample, while the other half consisted of the β-galactosidase substrate ONPG. Spectrophotometric measurements were started immediately after mixing. Unless otherwise mentioned, all values were subsequently normalized toward the total protein content per sample. Statistical Analysis. Statistical analysis was used to calculate and compare the data from IF and NIF sets. Since only comparisons between the mean values of small data sets were made, the Student t test was applied. Since we expected the IF sets to have at least an equal or larger mean value (βgalactosidase activity) than the NIF sets on the one hand, combined with the fact that a larger mean value for the NIF sets compared to the IF sets can only be attributed to chance, it was appropriate to choose a one-tail P-value. For all experiments, n ) 5, and the alpha level (R-level) was set at 0.05. All statistical analyses were carried out using GraphPad Prism Version 4.00 and on-line GraphPad statistical software (GraphPad Software, Inc., San Diego, CA). Miscellaneous Procedures. Protein concentrations were determined by the Bradford assay25 using the Bio-Rad Protein Assay Solution (Bio-Rad, Hercules, CA). SDS-PAGE was carried out according to standard protocols.26 Western blotting using an antibody specific for β-galactosidase from E. coli was performed according to standard protocols. Staining specific for heme-containing proteins was carried out as described by Thomas et al.27 Image acquisition, and image processing to make up Supporting Information Figure 2 were carried out using software packages Corel version 9 and Paintshop Pro version 5 (Corel, Berkshire, U.K.).

Results Outline of the Two-Hybrid System. Crucial in this bacterial two-hybrid system are two inactive but complementing β-galactosidase truncations. The demarcations of the ω-donor (referred to as ∆R) and the R-donor (referred to as ∆ω) are based on previous studies by Mohler and Blau.28 The ∆R β-galactosidase protein lacks an N-terminal fragment consisting of amino acids 11-41, while ∆ω lacks the entire C-terminal part starting from amino acid 789. For the validation of our system, these β-galactosidase fragments are both fused to two interacting or two non-interacting proteins, in what follows termed ‘IF’ (interacting fusion protein set) and ‘NIF’ (noninteracting fusion protein set), respectively. It is expected that β-galactosidase reconstitution depends on the protein-protein recognition between the two proteins of interest, fused Nterminally to the complementing β-galactosidase truncations and expressed in their native Gram-negative cellular compartment. Assay Design. To validate our system, we aimed to (i) reconfirm some already well-documented protein-protein interactions, (ii) probe for the transient interaction between two periplasmic redox-active R. capsulatus proteins, and (iii) test our system in a mutation-driven structure-function study via deletion mapping analysis. Eleven proteins were selected based on their relative interaction behavior and their cellular localization: AmyA, DnaE, DnaN, FimC, FecB, FhuA, AtpB, AtpE, and AtpF from E. coli, and two proteins from R. capsulatus which are expected to transiently interact,30 that is, cytochrome c peroxidase (CCP) and cytochrome c2. A list of these proteins and their functions is given in Table 2, whereas Supporting Information Figure 1 schematically shows the experimental design. Two (supposedly) interacting and two non-interacting proteins were chosen for each bacterial cell 2590

Journal of Proteome Research • Vol. 6, No. 7, 2007

Borloo et al.

compartment. The cytoplasmic proteins AmyA and DnaN were fused to ∆R, while DnaE was fused to ∆ω. DnaE and DnaN are subunits of the DNA polymerase III holoenzyme and are known to interact with each other,29 which is not the case for protein couple AmyA (an R-amylase) and DnaE. It has been established that this interaction between DnaE and DnaN is directed by two distinct patches, encompassing the residues 920-92423 and 1154-1159,22 the latter being located at the very C-terminus of the protein. DnaE1153, DnaE920/1153, and DnaE917 mutants lacking one or both of these patches were constructed as described in the Experimental Section and fused to ∆ω, after which deletion mapping analysis was carried out. Testing the system for periplasmic proteins involved separately fusing FimC and cytochrome c2 to the ∆R part of β-galactosidase, and FecB and CCP to ∆ω. The interaction between CCP and cytochrome c2 has been suggested previously, based on the observed electron transfer from cytochrome c2 to CCP,30 but no interaction is expected for the protein couple FecB and FimC, an iron transport protein and a type I pilus chaperone, respectively. Finally, to validate the technique for membrane proteins, AtpB and FhuA were fused to ∆ω, while AtpF and AtpE were fused to ∆R. FhuA is an outer membrane receptor mediating TonBdependent import of ferrichrome, while AtpB, AtpF and AtpE constitute subunits of the F0 sector of the E. coli ATP-synthase.31 A general scheme of the cloning strategy is given in Figure 1. Experimental. Correct synthesis of the fusion proteins was controlled by Western blotting (immunoblot) for all proteins/ chimeras and by heme staining for those carrying cytochrome c2 and CCP. The immunodetection using anti-β-galactosidase antibodies verified that ∆R and ∆ω, as well as the different chimeras, were properly synthesized at low, physiological, levels when isopropyl-β,D-thiogalactopyranoside (IPTG) was added to the growth medium (Supporting Information Figure 2). Heme staining shows successful heme attachment for both cytochrome c2 and cytochrome c peroxidase (CCP) after proper translocation to the periplasm (Supporting Information Figure 2). To validate the in vivo character of our bacterial two-hybrid system, we expressed ∆R and ∆ω separately, after which cytoplasmic fractions were mixed in equivalent quantities in a volume of 1 mL. No β-galactosidase activity could be detected, emphasizing the need for coexpression. The bacterial two-hybrid setup was assayed in lacZ-deficient E. coli MC1061 cells. β-Galactosidase complementation was tested by coexpression of the ∆R and ∆ω muteins, both lacking an N-terminally located protein of interest, which as expected18,28 yielded high activities (3530 ( 696 nmol/(min‚mg); Figure 2A), contrasting to the results published by Rossi and co-workers.11 As a control, pB2H∆R/pB2H∆ω co-transformants grown without IPTG were found to be β-galactosidase-negative, indicating that protein synthesis is strictly controlled by the tac promoter. In contrast, when these β-galactosidase truncations both carried cytoplasmic non-interacting proteins, thus, forming a NIF set, enzyme activity was almost completely abolished (1.52 ( 1.02 nmol/(min‚mg); Figure 2B), as was also the case when two differently localized NIF set fusion proteins were assayed (activity values in the order of 1 nmol/(min‚mg) were observed), thereby reflecting the high specificity and sensitivity of our technique. However, when a protein of interest was fused to ∆R (e.g., ∆RΩDnaN) and subsequently coexpressed with ∆ω, an activity level comparable to that resulting from ∆R and ∆ω coexpression was noted, whereas a lower β-galactosidase activity (244 ( 41.0 nmol/(min‚mg)) was

research articles

β-Galactosidase-Based Bacterial Two-Hybrid System

Figure 1. Cloning strategy for the vectors pB2H∆R and pB2H∆ω. The tac promoter was obtained through EcoRV-PstI restriction of vector pALTER Ex1 and cloned into vectors pACYCDuet-1 and pETDuet-1. Subsequently, β-galactosidase truncations ∆R and ∆ω were cloned into the BamHI-NcoI sites of pACYCDuet-1Ωtac and pETDuet-1Ωtac, generating pB2H∆R and pB2H∆ω, respectively. Proteins of interest can then be cloned into both vectors using restriction sites SphI, NotI, PstI, AscI, and BamHI exclusively. Endogenous restriction sites are indicated in black, exogenous sites are displayed in blue, and removed restriction sites are in red.

detected when ∆R was coexpressed with ∆ωΩDnaE (Figure 2A). This indicates that the coupling of a protein to either ∆R or ∆ω has different implications on the reconstitution of the enzyme. The actual validation of the technique can be seen in Figures 2B and 2D. The cytoplasmic and membranous IF sets generated nearly 20-fold higher β-galactosidase activities compared to the NIF sets, with P-values of 0.0031 and 0.00062, respectively. The periplasmic IF set, however, displayed a 50-fold higher activity value (51.4 ( 15.9 nmol/(min‚mg)) compared to the FecB/FimC NIF set (P ) 0.0013) (Figure 2C). These observations are in accordance with the expectations. The interaction between the R subunit (DnaE) of the E. coli DNA polymerase III and the β subunit (DnaN) has been recently characterized,22,23,32 revealing that an internal patch consisting of residues 920-924 and a C-terminal patch encompassing amino acids 1154-1159 are involved in the R-β interaction, the 920-924 patch being the most important.23 Therefore, we performed a deletion mapping analysis by applying DnaE1153, DnaE920/1153, and DnaE917 mutants (for nomenclature, see Table 2), parallel with negative (AmyA/ DnaE) and positive (DnaE/DnaN) controls. Tests with DnaN/ DnaE920/1153 and DnaN/DnaE917 yielded activity values of 4.41 ( 1.30 and 4.11 ( 0.91 nmol/(min‚mg), respectively. This indicates that no interaction occurred, which is in accordance with previous reports22,23,32 and also demonstrates the qualitative character of our method in such a deletion mapping analysis in that, when all interaction patches are removed, a NIF set comparable signal is observed. However, performing our two-hybrid analysis with DnaN/DnaE1153 gave an activity

value of 23.8 ( 3.24 nmol/(min‚mg) which, after statistical analysis, was found not to be substantially lower (P ) 0.29) than that from DnaN/DnaE (Figure 2B). This is in contrast with the expectations that we derived from the results of Dohrmann and co-workers,23 who found that deleting amino acids 11541159 from DnaE would result in a 3.8-fold reduction of β2 binding. The latter experiment is indicative of the rather qualitative (but not quantitative) character of our method in deletion mapping analyses.

Discussion Evaluation of the Method. The feature that inactive truncations of β-galactosidase, missing the R or ω part of the monomer, can be rescued by complementation18 has been known for decades and in our hands also yielded the highest enzyme activity. In contrast, N-terminal addition of a protein to both β-galactosidase fragments highly reduces enzyme complementation, the level of which is dependent on the protein-protein recognition between the non-β-galactosidase parts. The fact that a NIF set yielded no or hardly detectable activity is a prerequisite for our system and ensures its high specificity. Interestingly, when fusing a protein to either the N-terminus of ∆R or ∆ω and coexpressing it with the complementary β-galactosidase truncation, activity levels were measured that were similar and ∼13-fold lower, respectively, compared to when ∆R and ∆ω were coexpressed (Figure 2A). This indicates that ∆R and ∆ω both display a different tolerance toward the N-terminal linkage of proteins. Inspection of the three-dimensional structure of wild-type β-galactosidase informs us that both the N- and C-termini play crucial roles in Journal of Proteome Research • Vol. 6, No. 7, 2007 2591

research articles

Borloo et al.

Figure 2. Results and schematic representation of the experiments. Normalized β-galactosidase activities (in nmol/(min‚mg); mean ( standard deviation) from the control experiments (A), cytoplasmic (B), periplasmic (C), and membranous (D) IF and NIF sets, together with a schematic representation of their cellular localization.

the build-up of the quaternary structure and its stabilization, and that the N-terminus in particular is involved in the formation of the active interface.33 As deduced from the results shown in Figure 2A, coupling a protein to ∆ω has drastic 2592

Journal of Proteome Research • Vol. 6, No. 7, 2007

implications on the formation of the activation interface, whereas fusing the same to ∆R has not. In principle, it is possible that these differences in activity are attributed to different expression levels of the chimeras compared to that

β-Galactosidase-Based Bacterial Two-Hybrid System

of ∆R and ∆ω, but immunoblot analyses indicated that this is not the case (Supporting Information Figure 2). Another possibility may be of structural origin, in that the β-galactosidase fragment, carrying a bound protein, would be prevented in its correct folding and complementation with its fragment partner. Straightforward evidence for this possibility will require a thorough structural study, which is outside the scope of the present article. So far, we have little evidence for such structureinduced effect, since we observed no difference in complementation when using either the small 32-kDa AtpB or the rather large 138-kDa DnaE as proteins bound to ∆ω. If later it is found that structure-induced effects do matter with other proteins, then this problem might be countered by using a longer linker than the four amino acid linker sequence (GGSG) we used in the present work. Validation. The actual validation of our two-hybrid strategy was carried out by coexpressing cytoplasmic, periplasmic, and membranous IF (interacting fusion protein couple) and NIF (non-interacting fusion protein couple) sets. The observation that IF sets consistently resulted in 20- to 50-fold higher β-galactosidase activity values when compared to NIF sets (Figure 2) confirms our expectations that additional bilateral attraction or repulsion forces, due to interacting or noninteracting fusion proteins, respectively, dominate the process of β-galactosidase complementation in IF and NIF sets. These additional forces are in fact at the very basis of the two-hybrid system we here describe. Nevertheless, and as also observed for the DnaE-DnaN interaction, the values from the IF sets represent only a fraction of the activity values measured from ∆R and ∆ω complementation. Apart from the potential structural reasons, which have been discussed above, this small signal can in principle also be attributed to a widely documented and often encountered problem, namely, that the β-galactosidase parts, which complement each other, have been cleaved off from the fusion proteins at the junction with the protein of interest. However, we can conclude that no such cleavage occurred, based on our findings that (i) Western blot analyses did not reveal bands corresponding to the separate ∆R and ∆ω fragments (Supporting Information Figure 2), (ii) if the activity values of IF sets were the consequence of cleavage events, the NIF sets would also have yielded substantially higher β-galactosidase activity values, which we did not observe, and (iii) experiments with ∆RΩDnaN combined with either ∆ωΩDnaE1153 or ∆ωΩDnaE920/1153 yielded only NIF set comparable activity values in the latter case, even though their junctions with the ∆ω part are identical and would thus render them equally susceptible to such cleavage events. The values presented in Figure 2 are all normalized toward the total cellular protein content, indicating that a specific cell compartment preparation step is not mandatory for a successful application of the method. However, since we aimed to confirm that the IF and NIF sets are properly localized, it was necessary to isolate the cellular compartments prior to analysis. We thereby noted some additional advantages, that is, a significantly improved signal-to-noise ratio (i.e., IF to NIF ratio) and the possibility to check whether all β-galactosidase activity is situated in the proper cellular compartment. This was demonstrated in our experiments with the membranous IF and NIF sets. When cell lysate was applied, an IF to NIF ratio of ∼20 was noted, whereas performing the same experiment with membrane preparations yielded a ratio of ∼100. This confirmed that β-galactosidase activity was located in the membrane

research articles fraction exclusively, thus, ruling out the possibility of cytoplasmic or periplasmic β-galactosidase contamination as a result of improper localization of either one of the fusion proteins in the IF set. Locating the proteins of interest to the N-terminus of the chimeras enabled us to clone the genes of interest with their native signal (leader) sequence, which implies that the potential protein partners are targeted to their native compartment. A major drawback of this approach is that, when testing for protein-protein interactions between membrane proteins, one has to take into account that both β-galactosidase parts of the chimeras should be localized on the same side of the membrane. The importance of this issue is demonstrated by our experiments with the AtpB/AtpE NIF and the AtpB/AtpF IF sets. AtpB and AtpF localize their C-terminus to the cytoplasmic side of the membrane, while AtpE directs this region to the periplasmic face (right part of Figure 2D). Another concern may result from not knowing the localization of the β-galactosidase portion itself. This was revealed in an experiment where we performed the β-galactosidase activity assay with the AtpB/ AtpE and FhuA/AtpE NIF sets. We observed a ∼3-fold higher activity for the latter set (left part of Figure 2D). Both the outer membrane located FhuA and the inner membrane situated AtpE direct their β-galactosidase portion toward the periplasm, which illustrates that it is crucial to know the location of the β-galactosidase fragments in order to avoid false-positive results. In the past, studies have been dedicated to investigate whether β-galactosidase, being a cytoplasmic protein, is toxic in other cellular compartments. Snyder and Silhavy proposed it to be toxic when present in the bacterial periplasm since the full-length enzyme seems to assume a tightly folded conformation that cannot be secreted, presumably because of jamming the protein secretion machinery.34 In contrast, Freudl et al.35 have revealed that β-galactosidase can be transported across the cytoplasmic membrane when, as is the case with all expressed chimeras in our study, it is synthesized at low levels. Our experiments confirm this, since all chimeras that were directed to the inner membrane, outer membrane, or periplasm were properly localized and maturated (Supporting Information Figure 2). In summary, we have demonstrated that the β-galactosidase based bacterial two-hybrid system is an in vivo technique, allowing protein-protein interactions to be efficiently detected in their native cellular environment, that is, the cytoplasm, the periplasm, or the inner and outer membrane of Gram-negative bacterial cells. Moreover, aided by the fact that protein overexpression is not required, and even unwanted, the straightforward approach of picking up the gene from genomic DNA, cloning it into the proper vectors, and subsequently being able to analyze it for interactions, renders the technique nonlaborious and cost-efficient, paving the way toward its use in interactome research and proteome-scale screens. Furthermore, by incorporating the heme-containing redox-proteins cytochrome c2 and cytochrome c peroxidase in our study, we provided evidence that the system can also be applied to proteins that interact transiently and undergo post-translational modifications prior to exerting their cellular function. Moreover, our experiments with the DnaE mutants demonstrate the potential of the two-hybrid system to use it qualitatively in mutation-driven structure-function studies where the effect of well-chosen single or multiple mutations on the interaction behavior can be assessed. A drawback may consist of the Journal of Proteome Research • Vol. 6, No. 7, 2007 2593

research articles difficult expression of some proteins in E. coli and the inability of the bacterium to perform certain post-translational modifications, thereby limiting the number of eukaryotic protein couples that can be analyzed by our technique. Nevertheless, we believe that our two-hybrid system deserves a place among other frequently used techniques for the detection and identification of protein partners. Benefiting from the use of two different vectors, it may also be possible to detect protein oligomerization.

Acknowledgment. This work was supported by a personal grant to J.B. from the Institute for the Promotion of Innovation by Science and Technology in Flanders (IWTVlaanderen). J.J.V.B. and B.D. are indebted to the Fund for Scientific Research (FWO-Vlaanderen) for granting research project G.0190.04, as well as to the Bijzonder Onderzoeksfonds of Ghent University for Concerted Research Action GOA 120154. We thank Sarah De Keulenaer for technical help. Supporting Information Available: Supporting Figure 1, general experimental setup. Transformation schedule followed to obtain the IF and NIF E. coli MC1061 clones applied in this study. Supporting Figure 2, controls regarding the proper synthesis of the (fusion) proteins. (A) Western blot using antibody specifically against β-galactosidase. Equal amounts of protein were loaded (∼20 µg total protein). The molecular weight marker is indicated on the right. Lane 1, wildtype E. coli MC1061 cells; Lane 2, pB2H∆R (118 kDa); Lane 3, pB2H∆RΩamyA (177 kDa); Lane 4, pB2H∆RΩdnaN (162 kDa); Lane 5, pB2H∆ωΩdnaE (232 kDa); Lane 6, pB2H∆RΩfimC (147 kDa); Lane 7, pB2H∆ωΩfecB (129 kDa); Lane 8, pB2H∆ωΩfhuA (178 kDa); Lane 9, pB2H∆ωΩatpB (126 kDa); Lane 10, pB2H∆RΩatpE (128 kDa); Lane 11, pB2H∆RΩatpF (135 kDa); Lane 12, pB2H∆ω (94 kDa). The presented blot was merged from three original blots as indicated by vertical black lines using software as mentioned in Experimental Section. (B) Staining for heme proteins. Equal amounts of protein were loaded (∼20 µg total protein). The molecular weight marker is indicated on the left. Lane 1, pB2H∆RΩc2 (132 kDa); Lane 2, pB2H∆ωΩCCP (133 kDa). This material is available free of charge via the Internet at http:// pubs.acs.org. References (1) Wallrabe, H.; Periasamy, A. Imaging protein molecules using FRET and FLIM microscopy. Curr. Opin. Biotechnol. 2005, 16 (1), 1927. (2) Fields, S.; Song, O. A novel genetic system to detect proteinprotein interactions. Nature 1989, 340 (6230), 245-6. (3) Chien, C. T.; Bartel, P. L.; Sternglanz, R.; Fields, S. The two-hybrid system: a method to identify and clone genes for proteins that interact with a protein of interest. Proc. Natl. Acad. Sci. U.S.A. 1991, 88 (21), 9578-82. (4) Chong, J. A.; Mandel, G. In The Yeast Two-Hybrid System; Bartel, P. L., Fields, S., Eds.; Oxford University Press: New York, 1997; pp 289-97. (5) Zhang, B.; Kraemer, B.; Sengupta, D.; Fields, S.; Wickens, M. In The Yeast Two-Hybrid System; Bartel, P. L., Fields, S., Eds.; Oxford University Press: New York, 1997; pp 298-315. (6) Vidal, M.; Braun, P.; Chen, E.; Boeke, J. D.; Harlow, E. Genetic characterization of a mammalian protein-protein interaction domain by using a yeast reverse two-hybrid system. Proc. Natl. Acad. Sci. U.S.A. 1996, 93 (19), 10321-6. (7) Fromont-Racine, M.; Rain, J. C.; Legrain, P. Toward a functional analysis of the yeast genome through exhaustive two-hybrid screens. Nat. Genet. 1997, 16 (3), 277-82.

2594

Journal of Proteome Research • Vol. 6, No. 7, 2007

Borloo et al. (8) Uetz, P.; Giot, L.; Cagney, G.; Mansfield, T. A.; Judson, R. S.; Knight, J. R.; Lockshon, D.; Narayan, V.; Srinivasan, M.; Pochart, P.; Qureshi-Emili, A.; Li, Y.; Godwin, B.; Conover, D.; Kalbfleisch, T.; Vijayadamodar, G.; Yang, M.; Johnston, M.; Fields, S.; Rothberg, J. M. A comprehensive analysis of protein-protein interactions in Saccharomyces cerevisiae. Nature 2000, 403 (6770), 6237. (9) Michnick, S. W.; Remy, I.; Campbell-Valois, F. X.; Vallee-Belisle, A.; Pelletier, J. N. Detection of protein-protein interactions by protein fragment complementation strategies. Methods Enzymol. 2000, 328 208-30. (10) Pelletier, J. N.; Campbell-Valois, F. X.; Michnick, S. W. Oligomerization domain-directed reassembly of active dihydrofolate reductase from rationally designed fragments. Proc. Natl. Acad. Sci. U.S.A. 1998, 95 (21), 12141-6. (11) Rossi, F.; Charlton, C. A.; Blau, H. M. Monitoring protein-protein interactions in intact eukaryotic cells by beta-galactosidase complementation. Proc. Natl. Acad. Sci. U.S.A. 1997, 94 (16), 8405-10. (12) Hu, J. C.; Kornacker, M. G.; Hochschild, A. Escherichia coli one- and two-hybrid systems for the analysis and identification of protein-protein interactions. Methods 2000, 20 (1), 8094. (13) Hu, J. C.; O’Shea, E. K.; Kim, P. S.; Sauer, R. T. Sequence requirements for coiled-coils: analysis with lambda repressorGCN4 leucine zipper fusions. Science 1990, 250 (4986), 1400-3. (14) Hennecke, F.; Muller, A.; Meister, R.; Strelow, A.; Behrens, S. A ToxR-based two-hybrid system for the detection of periplasmic and cytoplasmic protein-protein interactions in Escherichia coli: minimal requirements for specific DNA binding and transcriptional activation. Protein Eng. Des. Sel. 2005, 18 (10), 477-86. (15) Karimova, G.; Pidoux, J.; Ullmann, A.; Ladant, D. A bacterial two-hybrid system based on a reconstituted signal transduction pathway. Proc. Natl. Acad. Sci. U.S.A. 1998, 95 (10), 5752-6. (16) Wehrman, T.; Kleaveland, B.; Her, J. H.; Balint, R. F.; Blau, H. M. Protein-protein interactions monitored in mammalian cells via complementation of beta -lactamase enzyme fragments. Proc. Natl. Acad. Sci. U.S.A. 2002, 99 (6), 3469-74. (17) Jacobson, R. H.; Zhang, X. J.; DuBose, R. F.; Matthews, B. W. Three-dimensional, structure of beta-galactosidase from E. coli. Nature 1994, 369 (6483), 761-6. (18) Ullmann, A.; Jacob, F.; Monod, J. Characterization by in vitro complementation of a peptide corresponding to an operatorproximal segment of the beta-galactosidase structural gene of Escherichia coli. J. Mol. Biol. 1967, 24 (2), 339-43. (19) Wehrman, T. S.; Casipit, C. L.; Gewertz, N. M.; Blau, H. M. Enzymatic detection of protein translocation. Nat. Methods 2005, 2 (7), 521-7. (20) Sambrook, J.; Fritsch, E. F.; Maniatis, T. Molecular Cloning: A Laboratory Manual, 2nd ed.; Cold Spring Harbor Laboratory Press: Cold Spring Harbor, NY, 1989. (21) De Smet, L.; Pettigrew, G. W.; Van Beeumen, J. J. Cloning, overproduction and characterization of cytochrome c peroxidase from the purple phototrophic bacterium Rhodobacter capsulatus. Eur. J. Biochem. 2001, 268 (24), 6559-68. (22) De Saro, F. J. L.; Georgescu, R. E.; O’Donnell, M. A peptide switch regulates DNA polymerase processivity. Proc. Natl. Acad. Sci. U.S.A. 2003, 100 (25), 14689-94. (23) Dohrmann, P. R.; McHenry, C. S. A bipartite polymeraseprocessivity factor interaction: only the internal beta binding site of the alpha subunit is required for processive replication by the DNA polymerase III holoenzyme. J. Mol. Biol. 2005, 350 (2), 22839. (24) Imlay, K. R.; Imlay, J. A. Cloning and analysis of SodC, encoding the copper-zinc superoxide dismutase of Escherichia coli. J. Bacteriol. 1996, 178, 2564-71. (25) Bradford, M. M. A rapid and sensitive method for the quantitation of microgram quantities of protein utilizing the principle of protein-dye binding. Anal. Biochem. 1976, 72, 248-54. (26) Laemmli, U. K. Cleavage of structural proteins during the assembly of the head of bacteriophage T4. Nature 1970, 227 (5259), 680-5. (27) Thomas, P. E.; Ryan, D.; Levin, W. An improved staining procedure for the detection of the peroxidase activity of cytochrome P-450 on sodium dodecyl sulfate polyacrylamide gels. Anal. Biochem. 1976, 75 (1), 168-76.

research articles

β-Galactosidase-Based Bacterial Two-Hybrid System (28) Mohler, W. A.; Blau, H. M. Gene expression and cell fusion analyzed by lacZ complementation in mammalian cells. Proc. Natl. Acad. Sci. U.S.A. 1996, 93 (22), 12423-7. (29) Kubawara, N.; Uchida, H. Functional cooperation of the dnaE and dnaN gene products in Escherichia coli. Proc. Natl. Acad. Sci. U.S.A. 1981, 78 (9), 5764-5767. (30) Koh, M.; Meyer, T. E.; De Smet, L.; Van, Beeumen, J. J.; Cusanovich, M. A. Characterization of the interaction of Rhodobacter capsulatus cytochrome c peroxidase with charge reversal mutants of cytochrome c(2). Arch. Biochem. Biophys. 2003, 410 (2), 230-7. (31) Zhou, Y.; Duncan, T. M.; Cross, R. L. Subunit rotation in Escherichia coli FoF1-ATP synthase during oxidative phosphorylation. Proc. Natl. Acad. Sci. U.S.A. 1997, 94 (20), 10583-7.

(32) Lamers, M. H.; Georgescu, R. E.; Lee, S. G.; O’Donnell, M.; Kuriyan, J. Crystal, structure of the catalytic alpha subunit of E. coli replicative DNA polymerase III. Cell 2006, 126 (5), 881-92. (33) Matthews, B. W. The structure of E. coli beta-galactosidase. C. R. Biol. 2005, 328 (6), 549-56. (34) Snyder, W. B.; Silhavy, T. J. Beta-galactosidase is inactivated by intermolecular disulfide bonds and is toxic when secreted to the periplasm of Escherichia coli. J. Bacteriol. 1995, 177 (4), 953-63. (35) Freudl, R.; Schwarz, H.; Kramps, S.; Hindennach, I.; Henning, U. Dihydrofolate, reductase (mouse) and beta-galactosidase (Escherichia, coli) can be translocated across the plasma membrane of E. coli. J. Biol. Chem. 1988, 263 (32), 17084-91.

PR070037J

Journal of Proteome Research • Vol. 6, No. 7, 2007 2595