Protein Engineering, Design & Selection vol. 24 no. 11 pp. 819–828, 2011 Published online September 6, 2011 doi:10.1093/protein/gzr043
Use of periplasmic target protein capture for phage display engineering of tight-binding protein – protein interactions Bartlomiej G.Fryszczyn 1,2, Nicholas G.Brown 1,2, Wanzhi Huang 2, Miriam A.Balderas 3 and Timothy Palzkill 1,2,4 1
Department of Biochemistry and Molecular Biology, Baylor College of Medicine, One Baylor Plaza, Houston, TX 77030, USA, 2Deparment of Pharmacology, Baylor College of Medicine, One Baylor Plaza, Houston, TX 77030, USA and 3Deparment of Molecular Virology and Microbiology, Baylor College of Medicine, One Baylor Plaza, Houston, TX 77030, USA
To whom correspondence should be addressed. E-mail: [email protected]
Received March 23, 2011; revised July 7, 2011; accepted August 5, 2011 Edited by Philipp Holliger
Phage display is a powerful tool to study and engineer protein and peptide interactions. It is not without its limitations, however, such as the requirement for target protein purification and immobilization in a correctly folded state. A protein capture method is described here that allows enrichment of tight-binding protein variants in vivo thereby eliminating the need for target protein purification and immobilization. The linkage of genotype to phenotype is achieved by placing both receptor and ligand encoding genes on the same plasmid. This allows the isolation of the tight-binding ligand –receptor pair complexes after their association in the bacterial periplasm. The interaction between the TEM-1-b-lactamase fused to the gene 3 coat protein displayed on the surface of M13 bacteriophage and the b-lactamse inhibitory protein (BLIP) expressed in soluble form with a signal sequence to export it to the periplasm was used as a model system to test the method. The system was experimentally validated using a previously characterized collection of BLIP alanine mutants with a range of binding affinities for TEM-1 b-lactamase and by isolating tight-binding variants from a library of mutants randomized at residue position Tyr50 in BLIP which contacts b-lactamase. Keywords: antibiotic resistance/b-lactamase/b-lactamase inhibitory protein/phage display/protein – protein interactions
Introduction The regulation of cellular processes is largely governed by protein:protein interactions. These interactions serve as targets for the development of treatments that disrupt protein complexes. The design and development of high affinity and specificity between interacting partners is an important goal
of protein engineering. Therefore, methods to engineer protein:protein interactions are critical tools for the development of new therapeutics and diagnostics. Phage display technology has proven to be a useful method for identifying and engineering protein:protein interactions (Levin and Weiss, 2006; Bradbury et al., 2011). For this purpose, a protein of interest is expressed as a fusion with either gene 3 protein ( pIII) or gene 8 protein ( pVIII) that are incorporated into the phage particle which encases the DNA encoding the fusion protein (Smith, 1985; Petrenko, 2008). Combinatorial libraries of protein variants can be created and used in a selection process against an immobilized target to identify variants that bind tightly to the target molecule (Levin and Weiss, 2006; Petrenko, 2008). While this method has proven to be useful, it requires the target protein to be purified and properly folded on immobilization. As a possible alternative, we present a system that enables the enrichment of tight-binding interaction mutants without the requirement of target purification or immobilization and occurs in the context of the Escherichia coli periplasmic space. The b-lactamase:b-lactamase inhibitory protein (BLIP) interaction is a well-established model system of protein:protein recognition and was used to test the feasibility of capture of the target protein in the periplasm by the pIII fusion as a method to enrich for tight-binding variants from a combinatorial library. b-Lactamases provide for bacterial antibiotic resistance by hydrolyzing b-lactam antibiotics that include the penicillins and cephalosporins (Bush and Jacoby, 2010). These enzymes are widely distributed in both Gram-positive and Gram-negative bacteria and are divided into four classes (A– D) based on primary amino acid sequence similarities (Ambler, 1980). The Class A b-lactamases utilize an active site serine residue for catalysis and, as a family, can hydrolyze a broad range of b-lactam antibiotics including penicillins, cephalosporins and, in some cases, carbapenems (Bush, 2010). The TEM-1 b-lactamase is a Class A enzyme that efficiently hydrolyzes penicillins and many cephalosporins and is the most common plasmid-encoded b-lactamase found in Gram-negative bacteria (Majiduddin et al., 2002). BLIP is a 17.5 kDa protein inhibitor of Class A b-lactamases and is produced from the soil bacterium Streptomyces clavuligerus (Doran et al., 1990). The interaction between BLIP and Class A b-lactamases has been studied extensively with several structures of BLIP – b-lactamase complexes available in addition to detailed kinetic and thermodynamic data describing the interactions (Strynadka et al., 1996; Zhang and Palzkill, 2003, 2004; Reynolds et al., 2006; Reichmann et al, 2007; Wang et al., 2007, 2009). BLIP inhibits Class A b-lactamases with a wide
# The Author 2011. Published by Oxford University Press. All rights reserved. For Permissions, please e-mail: [email protected]
B.G.Fryszczyn et al.
range of affinities with dissociation constants from picomolar to micromolar depending on the target and inhibits the TEM-1 enzyme with a Ki of 0.5 nM (Strynadka et al., 1996; Zhang and Palzkill, 2004). The TEM-1/BLIP interaction has been previously shown to be a suitable protein model for phage display studies in that TEM-1 has been demonstrated to be properly folded on the surface of the phage and to bind specifically and with high affinity to BLIP (Rudgers and Palzkill, 1999; Huang et al., 2000a,b). In addition, the affinity and specificity determinants in BLIP for binding to Class A b-lactamase have been identified through the use of alanine-scanning mutagenesis of residues in BLIP that contact TEM-1 b-lactamase based on the X-ray structure of the complex (Zhang and Palzkill, 2003, 2004). Determination of inhibition constants for the BLIP alanine mutant interaction with TEM-1 b-lactamase generated important information on residues critical for binding but also provided a set of mutants with a range of affinities for TEM-1 b-lactamase that were used in this study to validate the phage display system. Materials and methods
Bacterial strains The E.coli strain XL1-Blue [recA1, endA1, gyrA96, thi-1, hsdR17, supE44, relA1, lac, [F9 proAB lacIq lacZDM15, Tn10 (tetr)]] (Stratagene, Inc.) was used for cloning and for preparation of the initial, naive phages of the BLIP Y50X library (Bullock et al., 1987). Escherichia coli strain TG1 [F9 traD361, lacIq, 4(lacZ) M15 proA1B1/supE D(hsdM2mcrB2) (r2 m2 McrB2) thi 4(lac2proAB)] was used for subsequent phage titre determination and propagation during the rounds of phage display enrichment of binders. The BLIP mutant proteins of interest were expressed for purification using E.coli RB791 (strain W3110 lacIqL8) (Amann et al., 1983). Escherichia coli BL21(DE3) strain (F – 2 ompT gal dcm lon hsdSB(r2 B mB ) l(DE3 [lacI lacUV5-T7 gene 1 ind1 sam7 nin5])) was utilized for b-lactamase expression and purification (Studier and Moffatt, 1986).
Construction of phage display plasmids The pTP208 plasmid contains the blaTEM-1 gene fused at its 30 -end (C-terminus of enzyme) to 50 end of the gene encoding the C-terminal (CT) domain of gene 3 protein. This plasmid contains the blaTEM-1 constitutive promoter and blaTEM-1 signal sequence for secretion. It was constructed by converting the amber codon TAG that is present between blaTEM-1 and gene III in the previously described pTP145 phage display plasmid to a CTT leucine codon by sitedirected mutagenesis (Huang et al., 2000a). The pTP268 plasmid was constructed from the pTP208 plasmid by insertion of the gene encoding BLIP along with the bacterial trc promoter sequence (Brosius et al., 1985; Amann et al., 1988). This was accomplished by PCR amplification of the trc promoter-BLIP gene from the BLIP expression plasmid pGR32, which has been described previously (Fig. 1) (Petrosino et al., 1999). The oligonucleotide primers used for amplification were: pTP123-trc-NheI, 50 -GGGCGGCTAGC TTGCGCCGACATCATAACGGTTCTGGC-30 ; and ptrc-2, 50 -CTCTCATCCGCCAAAACAG-30 . The resultant PCR product DNA was purified and digested with the NheI and 820
Fig. 1. Schematic diagrams of the pTP208 and pTP268 phage display plasmids and pGR32 BLIP protein expression plasmid. Plasmid maps for pTP208, pTP268 and pGR32 are shown. Restriction sites used to construct the pTP268 plasmid and the BLIP alanine mutants in pTP268 are shown. Details of the plasmid constructions are described in Materials and methods.
HindIII restriction endonucleases as was the pTP208 plasmid. The restriction enzyme-digested PCR product and pTP208 plasmid were each purified by agarose gel electrophoresis and the DNA was further purified using a Qiagen gel purification kit. The NheI– HindIII digested PCR DNA fragment was inserted into the NheI– HindIII digested pTP208 plasmid using T4 DNA ligase to create pTP268. The sequence of the BLIP DNA insertion in pTP268 was verified by DNA sequencing. The BLIP alanine mutants were constructed previously in the pGR32 protein expression plasmid (Fig. 1) (Petrosino et al., 1999; Zhang and Palzkill, 2003). The BLIP genes with alanine mutations were introduced into the pTP268 plasmid by release of the BLIP gene from pGR32 by digestion with the EcoRI and XbaI restriction endonucleases and ligation of this fragment into pTP268 that was digested with the EcoRI and XbaI enzymes (Fig. 1). The result was replacement of the wild-type BLIP gene in pTP268 with each of the BLIP alanine mutants. DNA sequencing was used to verify the presence of the alanine mutation in each plasmid.
Single-point enzyme-linked immunosorbent assay (ELISA) experiments The pTP208 plasmid and the BLIP wild-type and alanine mutants in the pTP268 plasmid were transformed into E.coli TG1 cells by electroporation and spread on lysogeny broth (LB) agar plates containing 12.5 mg/ml chloramphenicol. Single colonies of each BLIP variant were used to inoculate 1 ml of 2YT medium containing 12.5 mg/ml chloramphenicol in a 2 ml 96-deep well bacterial growth plate. The cultures were grown by shaking the plate at 378C for 3 – 4 h and the cultures in each well were infected with 1010 KM13 helper phage particles and let to stand at 378C for 30 min. The 96-well culture plate was centrifuged at 3000 g for 15 min and the supernatant of each well was decanted. The cells in each well were resuspended in 1 ml 2YT medium
Phage display panning by periplasmic target protein capture
supplemented with 12.5 mg/ml chloramphenicol and 50 mg/ ml kanamycin and grown with shaking overnight at 378C. The 96-well culture plate was then centrifuged at 3000 g and the supernatant containing the phage particles was collected to be used for ELISA experiments. The ELISA experiments were initiated by adding either 0.1 ml of 90 nM anti-His-tag monoclonal antibody (Qiagen), or 90 nM of bovine serum albumin (BSA), or a 1:500 dilution of anti-TEM-1 polyclonal sera to wells of a 96-well immulon 2 high binding (HB) microtitre plate (Thermo Scientific) and left standing at 48C overnight to immobilize the proteins into the wells. The buffer used for the protein preparations was phosphate-buffered saline (PBS) pH 7.4. After immobilization, the plate was washed three times with 0.2 ml/well of PBS pH 7.4. The wells were blocked by gently shaking the plate at room temperature with 0.2 ml/ well of PBS (MPBS) (5% dry milk in PBS) for 2 h. PBS was again used to wash the wells three times and 0.1 ml/well of phage supernatants was added to the wells. The phage particles were incubated to bind to the target protein for 2 h with shaking at room temperature. The unbound phages were removed by washing the wells three times with 0.2 ml/well of PBST (PBS with 0.1% Tween20). A 1:3000 dilution of anti-M13-HRP (GE Healthcare) was added to wells at a volume of 0.1 ml/well. This solution was incubated for 1 h at room temperature and the plates were then washed with 0.2 ml/well of PBST four times and developed with 0.1 ml/ well of 3,30 ,5,50 -tetramethyl benzidine (TMB) peroxidase solution (KPL). The ELISA was read at 630 nm after 30– 45 min of peroxidase reaction with an EL800 universal microplate reader (BioTek). This ELISA protocol was also followed for characterizing tight-binding clones from the BLIP Y50X phage library selection experiments.
Construction of BLIP Y50X random library Amino acid position 50 of BLIP was previously randomized by site-directed mutagenesis to convert the TAC codon to NNS where N is any of the four nucleotides and S is C or G in the BLIP expression plasmid pGR32 and a collection of each of the 20 individual substitutions was generated (Yuan et al., 2009). Each of the BLIP genes with a Y50 substitution was introduced into the pTP268 phage display plasmid by digestion of pGR32-based plasmid with the XbaI and EcoR1 restriction enzymes and ligation of the resultant BLIP-containing fragment into XbaI– EcoRI-digested pTP268 plasmid using T4 DNA ligase. The ligation reaction mixtures were used to transform E.coli XL1-Blue cells by electroporation and transformants were spread on LB agar plates supplemented with 12.5 mg/ml chloramphenicol. Plasmid DNA was isolated and the DNA sequence of the BLIP gene was determined for position 50 to ensure that the expected codons were present at this position. The cells encoding each mutant were pooled and a phage preparation was made to create the BLIP Y50X library.
Preparation of M13 phages displaying TEM-1 b-lactamase M13 bacteriophage propagation and purification were performed as previously described (Huang et al., 2000a; Brown and Palzkill, 2010). The phages encoding the TEM-1 b-lactamase and wild-type BLIP and each of the BLIP alanine mutants were produced by transformation of the pTP268 plasmid for wild-type BLIP or the pTP268-derived
plasmids containing the BLIP alanine substitutions into E.coli TG1 cells by electroporation and colonies were selected on LB agar plates containing 12.5 mg/ml chloramphenicol. Single colonies of BLIP wild-type and each mutant were picked and grown overnight in 2 ml of 2YT broth containing 12.5 mg/ml chloramphenicol. The overnight cultures were used to inoculate 50 ml of 2YT broth supplemented with 12.5 mg/ml chloramphenicol. The cultures were grown to mid-log phase and were infected with 2 1011 KM13 helper phage particles. The culture was left standing at 378C for 30 min and phages were removed by centrifugation at 3000 g for 10 min. The supernatant was decanted and the E.coli TG1 pTP268-based cells were resuspended in 100 ml of 2YT broth containing 12.5 mg/ml chloramphenicol and 50 mg/ml kanamycin and grown overnight at 308C. The host E.coli TG1 cells harboring the pTP268 plasmid and pTP268-derived BLIP alanine mutant plasmids were pelleted by centrifugation at 3300 g for 30 min. The phage particles from each culture were precipitated from the supernatant with the addition of 20 ml 20% polyethylene glycol 6000– 2.5 M NaCl solution. This solution was left on ice for 1 h and centrifuged at 3300 g for 30 min. The supernatant was decanted and the pellet containing the phages was resuspended in 4 ml PBS. The remaining E.coli TG1 cells were removed by centrifugation at 11 600 g for 10 min. The titre of the phage preparations was determined by serial dilution of the phages into log-phase E.coli TG1 cultures. After a 30 min incubation, 0.1 ml of each dilution was spread on LB agar plates containing 12.5 mg/ml chloramphenicol and colonies were counted after overnight incubation at 378C to determine the phage titre. Phages from the BLIP Y50X library were prepared by inoculating 0.1 ml of pooled E.coli XL1-Blue colonies into 50 ml of 2YT broth supplemented with 12.5 mg/ml chloramphenicol. The remainder of purification process was as described above for the BLIP alanine mutants. After the initial selection of the BLIP Y50X phage (Round 1 biopanning), the enriched phages were propagated in E.coli TG1 cells.
Enrichment of BLIP mutants that bind tightly to TEM-1 b-lactamase The selection for tight-binding BLIP clones was performed by capturing the phages that were produced from E.coli TG1 containing the pTP268-derived plasmids with an anti-His-tag monoclonal antibody (Qiagen) which recognizes the 6X His-tag encoded on wild-type BLIP and BLIP mutants in pTP268. The antibody was immobilized onto Nunc MaxiSorp Immuno tubes by adding 2 ml of a 33 nM antibody solution into the tubes followed by overnight incubation at 48C. The immunotubes were then washed three times with PBS and blocked with 4 mls of PBS with 2% dry milk for 2 h at room temperature with gentle shaking. The tubes were washed three times with PBS and 2 ml of the phage preparations of interest were added to the tubes at a concentration of 1 1011 phages/ml. The phages were incubated in the immunotubes for 2 h at room temperature with gentle shaking to allow for the phages with BLIP bound to the displayed b-lactamase to interact with the anti-His-tag monoclonal antibody. The tubes were then washed 10 times with PBS containing 0.05% Tween 20 (PBST) to remove the unbound phages. The phages were eluted by adding 0.5 ml 821
B.G.Fryszczyn et al.
of 0.2 M glycine pH 2.0 with 1 mg/ml BSA for 8 min while shaking and rotating. It has previously been shown that pH 2.0 disrupts the interaction between BLIP and TEM-1 b-lactamase (Albeck and Schreiber, 1999). The low pH elution is also expected to disrupt the interaction between the His-tag on BLIP and the anti-His-tag antibody. The effectiveness of the low pH elution was directly demonstrated in by phage ELISA of the wild-type complex bound to His-tag antibody with and without low pH treatment (Supplementary Fig. S1). After elution, the phage solution was added to a microcentrifuge tube and neutralized with 0.15 ml of 1 M Tris pH 9.1. The titre of the phage elution solution was determined as described above. These phages were used to infect 1.25 ml of mid-log E.coli TG1 cells, which were then spread on chloramphenicol containing plates, and incubated overnight at 378C. Cells from the resulting bacterial lawn were then pooled and propagated in a 50 ml liquid culture (2YT media) for another round of selection. An identical procedure was performed in parallel for each phage preparation with a 1:500 dilution of anti-TEM-1 antibody as the immobilized target in place of the anti-His-tag antibody as a control to examine the population of phages enriched in the absence of selection for BLIP binding.
Enrichment of target BLIP variant from an excess of wild-type clones In order to determine the ability of the phage display system to identify rare tight-binding mutants from a library, mixtures of a tight-binding variant of BLIP (Y50A) and wild-type BLIP in ratios of 1:103 and 1:106 were created. The mixtures were then subjected to enrichment in order to determine the minimum number of rounds necessary to identify the tightbinding Y50A variant. The enrichment experiments were conducted identically to those described above for BLIP alanine and Y50X collections. Two separate cultures of E.coli TG1 cells expressing pTP268 containing BLIP Y50A and wild-type BLIP were grown to mid-log phase [optical density (OD)0.5]. Cells from both cultures were pooled in 1:103 and 1:106 ratios (Y50A in excess of WT) in 50 ml 2YT media and grown to mid-log phase. The cultures were then infected with 2 1011 KM13 helper phage particles. The cultures were left standing at 378C for 30 min and additional 50 ml of 2YT broth containing 12.5 mg/ml chloramphenicol was added and grown overnight at 308C. The pooled host E.coli TG1 cells harboring the pTP268 plasmid (WT BLIP) and pTP268-derived Y50A BLIP mutant plasmids were pelleted by centrifugation at 3300 g for 30 min and the phage were isolated as described above. The titre of the pooled phage preparations were determined by serial dilution of the phages into log-phase E.coli TG1 cultures as described above. The enrichment for the tight-binding Y50A BLIP clone was performed by capturing the phages with the anti-His-tag monoclonal antibody using the same experimental procedures as described above for the alanine mutants and the Y50X library. Three rounds of enrichment were performed for both the 1:103 and 1:106 mixtures and the colonies resulting from infection of E.coli TG1 with the elution mixtures were pooled and phages were generated for the next round of enrichment using helper phage as described above. The identical procedure was performed in parallel for the Y50A/wildtype BLIP pool with a 1:500 dilution of anti-TEM-1 822
antibody as the immobilized target in place of the anti-His-tag antibody as a control to examine the population of phages enriched in the absence of selection for BLIP binding. BLIP variants were selected at random for sequencing from each round of panning with either anti-His-tag monoclonal antibody or anti-TEM-1-b-lactamase polyclonal antibody as the immobilized target protein.
Test of the efficiency of low-pH elution The efficiency of elution of phages that were bound to the anti-His-tag antibody in the microtitre wells during enrichment was evaluated by phage ELISA The ELISA experiments were initiated by adding 0.1 ml of 90 nM anti-His-tag monoclonal antibody (Qiagen), or 90 nM of BSA to wells of a 96-well immulon 2 HB microtitre plate (Thermo Scientific) and left standing at 48C overnight to immobilize the proteins into the wells. The buffer used for the protein preparations was PBS pH 7.4. After immobilization, the plate was washed three times with 0.2 ml/well of PBS pH 7.4. The wells were blocked by gently shaking the plate at room temperature with 0.2 ml/well of MPBS (5% dry milk in PBS) for 2 h. PBS was again used to wash the wells three times and 0.1 ml/well of phage supernatants was added to the wells. The phage particles were incubated to bind to the target protein for 2 h with shaking at room temperature. The unbound phages were removed by washing the wells three times with 0.2 ml/well of PBST (PBS with 0.1% Tween20). A 1:3000 dilution of anti-M13-HRP (GE Healthcare) was added to wells at a volume of 0.1 ml/well. This solution was incubated for 1 h at room temperature and the plates were washed with 0.2 ml/ well of PBST 10 times. The elution efficiency was determined by eluting the phages either by adding 0.5 ml of 0.2 M glycine pH 2.0 with 1 mg/ml BSA or PBS pH 7.4 for 8 min while shaking and rotating. The tubes were washed three times with PBS and the ELISA was then developed with 0.1 ml/well of TMB peroxidase solution (KPL). Wells were read at 630 nm after 30– 45 min of peroxidase reaction with an Infinite Pro 2000 microplate reader (Tecan).
BLIP expression and purification A protein expression plasmid ( pGR32) for production of BLIP in E.coli was developed previously and wild-type BLIP as well as the BLIP alanine mutants have previously been constructed, expressed and purified using this system (Petrosino et al., 1999; Zhang and Palzkill, 2003, 2004). The BLIP expression plasmid ( pTP123) was introduced into E.coli RB791 and cells were grown in LB medium containing 12.5 mg/ml chloramphenicol at 378C and induced with 3 mM isopropyl b-D-1-thiogalactopyranoside (IPTG) at a culture density of A600 nm 1.2 and incubated for an additional 6 h at 238C. The culture was then harvested at 8000 r.p.m. for 15 min at 48C. Bacterial pellets were resuspended in bacterial protein extraction solution (1% Triton-X 100 in 10 mM Tris-HCl, pH 8.0, 150 mM NaCl) at a volume of 20 ml per 1 g of bacterial pellet. The resuspended cells were shaken vigorously for 30min at room temperature for lysis. The lysate was centrifuged at 30 000 g for 30min and the supernatant was mixed with Talon cobalt His-tag affinity resin for 2 h. BLIP wild-type and substituted variants were eluted from Talon cobalt resin using 200 mM imidazole in Tris-buffered saline and the purity of fractions was assessed by SDS-PAGE.
Phage display panning by periplasmic target protein capture
b-Lactamase inhibition assay The inhibition assay for evaluating the binding constant for tight-binding inhibitors has been described previously (Petrosino et al., 1999; Brown and Palzkill, 2010). Various concentrations of BLIP (0.03 – 1 nM for mutants Y50A, 0.03 – 1.5 nM for Y50 M and 0.09– 4 nM for wild type) were incubated with 0.3 nM of TEM-1 b-lactamase for 1 h at room temperature. The percentage of b-lactamase bound by BLIP was determined by monitoring the initial velocity of TEM-1-b-lactamase-mediated nitrocefin (chromogenic cephalosporin analog) hydrolysis in a spectrophotometer at OD485. A nitrocefin concentration of 40 mM was used in the binding assays. The Ki values for each BLIP mutant were determined by fitting the initial velocities to Morrison tightbinding equation (Morrison, 1969): Efree ¼ ½E0 ½E0 þ ½I0 þ Kiapp
qﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ ð½E0 þ ½I0 þ Kiapp Þ2 ð4½E0 ½I0 Þ 2 ð1Þ
where Efree is the concentration of free enzyme determined by residual activity of TEM-1 b-lactamase against the activity and concentration of the uninhibited b-lactamase, [E0] is the total enzyme concentration and [I0] is the total inhibitor concentration. Results
Construction of a phage display plasmid for periplasmic target protein capture The interaction of BLIP with TEM-1 b-lactamase was used as a model to develop and assess a phage display system for target protein capture in the E.coli periplasm. For this purpose, a plasmid for expression of TEM-1-b-lactamase fused to pIII was constructed. The gene encoding TEM-1 was fused at its 30 end to the 50 end of the gene section encoding the CT domain of pIII to create plasmid pTP208 (Fig. 1). The TEM-1-pIII-CT fusion is under the transcriptional control of the constitutive b-lactamase promoter in pTP208 (Fig. 1). The BLIP gene was placed on the same plasmid to ensure expression of the TEM-1-pIII-CT fusion and BLIP in the same cell to facilitate the direct connection between the genotype of the BLIP variant with the phenotype (binding to b-lactamase) (Fig. 1). Both the TEM-1-pIII-CT fusion and BLIP contain signal sequences for export to the periplasm for assembly of the TEM-1-pIII-CT protein into the phage particle and for the interaction of BLIP and the fusion protein on the surface of the phage as it transits the periplasm (Fig. 2). In addition, the BLIP protein contains a 6X His-tag sequence and is expressed under the transcriptional control of the IPTG-inducible Trc promoter (Amann et al., 1983). The plasmid construct containing these components was named pTP268 (Fig. 1). The rationale for enriching tight-binding mutants with this system is that BLIP mutants with high affinity will bind TEM-1 b-lactamase in the periplasm of Ecoli and be extruded with the phage particles. The phages are then sorted by using an antibody that recognizes BLIP to capture BLIP from the phage preparation. If the BLIP mutant
Fig. 2. Overview of the in vivo periplasmic protein capture selection. The pTP268 plasmid containsTEM-1-b-lactamase fused to the CT domain of the M13 gene III protein. The plasmid also encodes the gene for BLIP with a His-tag sequence. Both BLIP and the TEM-1-pIII-CT fusion protein are expressed in the bacterial cytoplasm and exported to the periplasmic space where they associate and are extruded with the phage particle.
binds tightly to TEM-1 b-lactamase, capture of BLIP will also capture the phage that is displaying b-lactamase and carrying the pTP268 plasmid encoding the BLIP mutant and the TEM-1-pIII-CT fusion. BLIP mutant proteins that bind weakly to b-lactamase will also be captured by the antibody but they will not be associated with a phage particle and therefore phages encoding weak-binding BLIP mutants will not be enriched (Fig. 2). Test of phage capture using alanine-scanning mutants of BLIP. As described above, the BLIP amino acid residues that are important for binding TEM-1 b-lactamase were previously identified using alanine-scanning mutagenesis of BLIP followed by the determination of inhibition constants for each BLIP alanine mutant (Petrosino et al., 1999; Zhang and Palzkill, 2003, 2004) (Fig. 3). A total of 15 of these BLIP alanine mutants with a range of affinities for TEM-1 b-lactamase were chosen to assess the binding affinity required for BLIP to remain associated with the phage particle during panning with a BLIP-specific antibody. The BLIP construct in the pTP268 vector used for phage display contains a His-tag sequence that has been shown to be effective for affinity chromatography and thus should be accessible for capture with an anti-His-tag monoclonal antibody (Zhang and Palzkill, 2003, 2004). Therefore, an anti-His-tag monoclonal antibody was immobilized in wells of a 96-well 823
B.G.Fryszczyn et al.
plate and phage preparations from E.coli containing the pTP268 plasmid encoding wild-type BLIP as well as pTP268-derived plasmids encoding the various BLIP alanine mutants were added to the wells and the retention of phage particles by BLIP was evaluated by ELISA with an anti-M13 antibody (Fig. 4). A positive control experiment for phage capture was performed in which a polyclonal anti-TEM-1 b-lactamase was immobilized in place of the anti-His-tag
Fig. 3. Structure of BLIP with studied residues highlighted. Residues colored red represent positions where alanine substitutions were constructed. An alanine substitution at position 50 (highlighted yellow) increases the binding affinity for TEM-1 b-lactamase by more than 50-fold.
antibody and a negative control was performed in which BSA was immobilized rather an antibody. The ELISA results in Fig. 4 indicate that BLIP alanine mutants known to have affinity for TEM-1 b-lactamase equal to or tighter than the wild type (Ki ¼ 0.5 nM), including BLIP Y50A (Ki ¼ 0.01 nM), Y51A (Ki ¼ 0.5 nM), S113A (Ki ¼ 0.1 nM) and F143A (Ki ¼ 0.6 nM) exhibited high ELISA signals and clearly bound and retained phage particles in the ELISA wells. In addition, the BLIP E31A (Ki ¼ 2.0 nM) mutant displayed an ELISA signal somewhat less than wild type but significantly above background levels (Fig. 4). In contrast, the remainder of the BLIP alanine mutants, with Ki values ranging from 11 to 784 nM, did not display ELISA signals significantly above the negative control value. The BLIP E73A (Ki ¼ 0.4 nM) was an exception to the trend as it exhibited a lower than expected signal in the ELISA experiments based on the Ki value (Fig. 4). A possible explanation is that isothermal titration calorimetry data suggest that the BLIP E73A binding constant is well above the value for wild-type BLIP, suggesting the mutant binds weakly to TEM-1 b-lactamase which is consistent with the ELISA results in Fig. 4 (Wang et al., 2007). Therefore, taken together the ELISA experiments indicate that BLIP mutants with a Ki value of 2 nM or less are able to associate with the TEM-1-b-lactamase displayed phage particle in the bacterial periplasm and remain bound to the phage after extrusion from the bacteria. Selection of tight-binding BLIP variants from the pooled alanine-scanning mutants. The above ELISA experiments demonstrate that tight-binding BLIP mutants can bind b-lactamase displaying phage in bacterial cells but this does not necessarily indicate the system can be used to isolate
Fig. 4. ELISA of BLIP alanine mutant collection expressed from the pTP268 phage display plasmid. ELISA results are shown for addition of phage mutant preparations to wells coated with anti-His-tag antibody which binds to the His-tag present on BLIP. The detection antibody recognizes M13 gene 8 protein and therefore indicates the presence of phage particles in the well. The retention of phage is due to binding between BLIP and the phage-displayed TEM-1 b-lactamase. The inhibition constants for each BLIP alanine mutant for binding TEM-1 b-lactamase as determined in a previous study are listed in parentheses (nM) (Zhang and Palzkill, 2003).
Phage display panning by periplasmic target protein capture
tight binders from libraries of mutants. For example, it is possible that BLIP dissociates from the b-lactamase in the phage preparations but then rebinds. This would not affect experiments with a single clone in the phage preparation but would disrupt the linkage between the binding phenotype and the genotype of the mutants in a library if dissociation and rebinding events occur in the purified phage preparation and result in a shuffling of the partnerships of BLIP mutants with phages. The possibility of identifying tight-binding mutants from a library of variants was tested by pooling the 15 alanine mutant variants into a single-phage preparation and enriching for tight binders with three rounds of panning of phages using the anti-His-tag monoclonal antibody as the immobilized target as described above for the ELISA experiments. As a control, TEM-1-b-lactamase polyclonal antibody was also used as an immobilized target. As b-lactamase is displayed on the phage surface, panning with this antibody does not select for BLIP-binding affinity and therefore should not enrich for a particular BLIP mutant. A total of 10 clones were sequenced after each round of panning with both the anti-His-tag and anti-b-lactamase antibodies to monitor the enrichment process (Table I). After one round of panning, the tightest-binding BLIP alanine mutant, Y50A (Ki ¼ 0.01 nM), was present in 4 of the 10 sequences and after three rounds it was present in 9 of 10 sequences. In contrast, there was no strong enrichment of any single BLIP alanine mutant in the b-lactamase antibody panning experiments. These results indicate that tightbinding BLIP mutants remain bound to TEM-1 b-lactamase on the phage surface in the pooled library and retain the genotype – phenotype linkage that allows for enrichment of the tightest-binding mutant by panning for the presence of BLIP. Selection of tight-binding mutants from a BLIP Y50X library. The periplasmic protein capture system was next used to probe the sequence requirements for position Y50 in BLIP for binding TEM-1 b-lactamase. It is known that the BLIP Y50A mutant binds 50-fold tighter than wild type to TEM-1 (Zhang and Palzkill, 2003). In addition, we
Table I. Results of biopanning of pooled BLIP alanine mutants expressed from the pTP268 phage display plasmid
previously used a genetic screen to show that many substitutions at position 50 function at levels similar to wild type (Yuan et al., 2009). The genetic system, however, is not sensitive enough to detect BLIP mutants that bind b-lactamase tighter than the wild type. The BLIP Y50X library was constructed by pooling equal amounts of cells containing pTP268 plasmid DNA encoding wild type and each of the 19 previously constructed Y50X variants and preparing phage from the pooled cells. This phage library was used for three rounds of panning on immobilized anti-His-tag monoclonal antibody to enrich for tight-binding mutants (Table II). As described above, the library was also subjected to three rounds of panning on immobilized anti-TEM-1-b-lactamase polyclonal antibody, which should not enrich for a particular BLIP mutant. A total of 10 clones were sequenced after each round of panning on each of the antibodies to monitor the extent of enrichment of BLIP mutants (Table II). As expected, there was no enrichment for a particular BLIP mutant from the panning experiments using anti-TEM-1-b-lactamase antibody. In contrast, there was an enrichment of the BLIP Y50A and Y50M mutants from the panning experiments using the anti-His-tag monoclonal antibody (Table II). The panning and sequencing results for the BLIP Y50X library suggest that the BLIP Y50M mutant, in addition to Y50A, binds to TEM-1 b-lactamase tighter than wild-type BLIP. This prediction was tested by purifying the BLIP wild type, Y50A and Y50M mutant proteins and measuring the inhibition constant (Ki) for TEM-1 b-lactamase (Materials and methods) (Fig. 5, Table III). It was found that the BLIP Y50M mutant was an 3-fold more potent inhibitor of TEM-1 b-lactamase than wild-type BLIP, which is consistent with the enrichment of this mutant as a tight binder in the panning experiment. Enrichment of the BLIP Y50A variant from an excess of wildtype clones. The value of phage display lies in the ability to identify variants with altered binding properties. In order to identify such variants, it is often necessary to enrich the mutants from a large, diverse library of combinatorial
Table II. Results of biopanning of pooled BLIP position 50 random library (Y50X) expressed from the pTP268 phage display plasmid
BLIP alanine mutant panninga BLIP Y50X panninga Round 1
Round 3 Round 1
Y50A Y50A Y50A Y51A F142A F142A F142A K74A W150A
W150A E31A F142A F142A R160A W112A K74A WT Y143A
Y50A Y50A Y50A Y50A Y50A Y51A Y142A W112A W112A
F142A W112A K74A W150A WT E73A K74A H41A R160A
Y50A Y50A Y50A Y50A Y50A Y50A Y50A Y50A Y51A
Anti-TEM Y143A/ W150A F142A W112A R160A K74A F142A Y51A W150A F142A E73A
The table lists 10 BLIP mutants selected at random from each round of panning with either anti-His-tag monoclonal antibody or ant-TEM-1-b-lactamase polyclonal antibody as the immobilized target protein.
Anti-His Y50A Y50A Y50A Y50A Y50M Y50S Y50S Y50E Y50Q Y50P
Round 2 Anti-TEM Y50N Y50Q Y50E Y50Q Y50H Y50T Y50S Y50Q Y50E Y50N
Anti-His Y50A Y50A Y50A Y50A Y50M Y50M Y50N Y50T Y50S Y50E
Round 3 Anti-TEM Y50N Y50E Y50H Y50Q Y50S Y50Q Y50M Y50N Y50M Y50S
Anti-His Y50A Y50A Y50A Y50M Y50M Y50M Y50M Y50E Y50S Y50Q
Anti-TEM Y50P Y50I Y50P Y50S Y50C Y50W Y50Q Y50E Y50T Y50G
The table lists 10 BLIP mutants selected at random from each round of panning with either anti-His-tag monoclonal antibody or anti-TEM-1-b-lactamase polyclonal antibody as the immobilized target protein.
B.G.Fryszczyn et al.
Table IV. Results of biopanning of a mixture of Y50A/BLIP and wt/BLIP clones in a 1:1000 ratio expressed from the pTP268 phage display plasmid Panning results from 1/1000 Y50A/WT mixturea Round 1
Fig. 5. Determination of BLIP WT, Y50A and Y50M Ki values for TEM-1 b-lactamase. BLIP inhibitory activity is expressed as the percentage of the remaining concentration of free b-lactamase at varying inhibitor concentrations (BLIP concentrations (nM) are shown on the x-axis). The lines represent the non-linear regression fit of the data to calculate the Ki (see Materials and methods). Each data point represents three independent experiments.
Table III. Ki values for the interaction of BLIP WT, Y50A and Y50M with TEM-1 b-lactamase BLIP
Anti-His WT WT WT WT WT WT WT WT WT WT
Round 2 Anti-TEM WT WT WT WT WT WT WT WT WT WT
Anti-His Y50A WT WT WT WT WT WT WT WT WT
Round 3 Anti-TEM WT WT WT WT WT WT WT WT WT WT
Anti-His Y50A Y50A Y50A WT WT WT WT WT WT WT
Anti-TEM WT WT WT WT WT WT WT WT WT WT
a The table lists BLIP variants selected at random from each round of panning with either anti-His-tag monoclonal antibody or anti-TEM-1-b-lactamase polyclonal antibody as the immobilized target protein. Ten clones were selected at random from Rounds 1– 3.
Table V. Results of biopanning of a mixture of Y50A/BLIP and wt/BLIP clones in a 1:1106 ratio expressed from the pTP268 phage display plasmid.
Ki (nM) Panning results from 1:1 106 Y50A/WT mixturea.
WT Y50A Y50M
0.40 + 0.05 0.07 + 0.01 0.12 + 0.01
Round 1 Anti-His 10 WT
substitutions. The periplasmic capture system was tested for the capacity to enrich a rare variant by using a starting mixture of the tight-binding BLIP Y50A mutant and wildtype BLIP in ratios of 1:103 and 1:106. Three rounds of biopanning were performed with immobilized anti-His-tag or anti-TEM-1 antibody as described for the alanine scanning and BLIP Y50A libraries. For the 1:103 mixture, 10 clones from each round of enrichment were sequenced and the BLIP Y50A was found in 10% of Round 2 and 30% of Round 3 clones from the anti-His-tag antibody panning and was not found among the anti-TEM-1 panning clones (Table IV). The tight-binding BLIP Y50A variant was not identified in the 1:106 mixture among Rounds 1 and 2 clones while 2 out of 40 randomly selected clones were BLIP Y50A after three rounds of panning on anti-His-tag antibody (Table V). The BLIP Y50A mutant was not identified in Rounds 1 – 3 clones from the anti-TEM-1 antibody panning (Table V). These results support the use of the system to significantly enrich tight-binding mutants from a large collection of other binding variants. Discussion Phage display was developed as a system to present peptides on the surface of a bacteriophage capable of infecting E.coli (Smith, 1985). Filamentous phage M13 is able to assemble and replicate without killing the bacterium and is the most commonly used bacteriophage for display systems (Levin and Weiss, 2006). Phage display is an important tool for studying protein:protein interactions because it facilitates the selection of peptides and proteins of desired functions and 826
Round 2 Anti-TEM 10 WT
Anti-His 10 WT
Round 3 Anti-TEM 10 WT
Anti-His 38 WT 2 Y50A
Anti-TEM 10 WT
a The table lists BLIP variants selected at random from each round of panning with either anti-His-tag monoclonal antibody or anti-TEM-1-b-lactamase polyclonal antibody as the immobilized target protein. Ten clones were selected at random from Rounds 1 and 2 and 40 random clones were selected from Round 3.
properties from molecular libraries (Smith, 1985; Sidhu and Koide, 2007). Most implementations of phage display, however, utilize a purified target protein which must be stable, and properly folded when immobilized. The system described here allows the interaction between the target protein and phagedisplayed receptor protein to occur in the periplasmic space and tight-binding mutants are subsequently selected from the extruded phages using a commercially available anti-His-tag antibody to capture the target protein and the associated phage particle. Although a His-tag antibody was used here, other affinity tag2antibody combinations would likely function equally well. In this study, the combinatorial library of mutants was present in the monomeric, soluble BLIP, while the wild-type TEM-1 was fused to the phage. In principle, however, either the soluble, non-fusion protein or the phagedisplayed protein could contain the combinatorial library of mutants. As the plasmid encodes both the receptor and the ligand genes it is possible to create combinatorial libraries in both proteins simultaneously, and look for tighter-binding variant combinations. This type of co-evolution experiment is not possible with standard phage display because the immobilized target protein is not encoded within the phage display system.
Phage display panning by periplasmic target protein capture
A phage display system that also utilizes expression of the target protein in the periplasm of E.coli for interaction with the phage-displayed protein has been described (de Wildt et al., 2002). This strategy is based on selection by avidity capture and makes use of the greatly increased affinity and half-life of multivalent compared with monovalent interactions (de Wildt et al., 2002). The target protein is fused to glutathione-S-transferase to induce dimerization and the receptor is displayed on the phage in multimeric form to allow a multivalent, long-lived, avidity interaction between the target and receptor (de Wildt et al., 2002). In contrast, the periplasmic target protein interaction with the phagedisplayed receptor described here utilizes a monovalent interaction. Phage ELISA experiments with BLIP alanine-scanning mutants indicated that a Kd value in the range of ,2 nM is required for BLIP to remain associated with the phage particle and bind immobilized anti-His-tag antibody. Panning of the BLIP Y50X library indicated that mutants with Kd values of 0.07 and 0.12 nM (Y50A and Y50M) could be enriched from a pool of mutants. Thus, the system works optimally for tight-binding protein – protein interactions. Selectively infective phage (SIP) technology also utilizes an interaction between a soluble target protein and a receptor protein displayed on the phage particle (Duenas and Borrebaeck, 1994; Gramatikoff et al., 1994; Jung et al., 1999). SIP directly couples a productive interaction between receptor and target protein with phage amplification and replication without the need for elution from a solid matrix. A target protein of interest is fused to the pIII N1 – N2 domains secreted to the periplasm and the receptor protein is fused to the CT domain of pIII. If the target and receptor protein interact, the pIII N1 – N2 domains are connected to the phage particle via the phage-displayed receptor-CT fusion and the extruded phage particles are able to infect E.coli. If the target and receptor proteins do not interact, the pIII CT domain alone cannot mediate infection and the phage cannot be amplified (Duenas and Borrebaeck, 1994; Gramatikoff et al., 1994; Jung et al., 1999). This system is also useful for tight-binding interactions but reconstitution of phage infectivity is inefficient and the target-receptor Kd binding requirements have not been examined systematically (Jung et al., 1999). The key to enrichment of tight-binding target protein mutants using the periplasmic capture system described here is likely to be a slow off-rate for the interacting proteins. The BLIP –TEM-1 b-lactamase interaction has been extensively studied and an off-rate of 1.6 1024 and 1.3 1024 s21 for BLIP dissociating from immobilized TEM-1 have been reported based on surface plasmon resonance and activitybased experiments have been reported (Albeck and Schreiber, 1999; Wang et al., 2009). The half-lives of the dissociation reaction based on these off-rates are 72 and 89 min, respectively. Therefore, it is possible that some dissociation and reassortment could occur within the phage preparation within the timeframe of the experiment. The high local concentration of the cognate phage-displayed receptor protein may drive rebinding events that avoid shuffling of the BLIP mutants with the TEM-1-b-lactamase phage. It is also possible that the predominance of the BLIP Y50A and Y50M mutants relative to wild-type BLIP in the combinatorial library selection in Table II is due to an even slower
off-rate in these tight-binding mutants which preserves the genotype – phenotype association and allows for the enrichment of these mutants. Therefore, it should be possible to use this system to engineer very-tight-binding, subnanomolar Kd, interactions. Phage display variants with altered binding properties are often selected from large, diverse libraries of mutants. As a further proof of concept the tight-binding BLIP variant Y50A was enriched from 1:103 and 1:106 mixtures with wild-type BLIP. Two rounds of enrichment were necessary to isolate the tight-binding BLIP Y50A variant from the 1:103 starting mixture while three rounds were needed to identify Y50A in 2 out of 40 clones sequenced using the 1:106 mixture (Tables IV and V). These results further support use of the method to identify tight-binding mutants in that the BLIP Y50A variant exhibits a Ki of 70 pM for inhibition of TEM-1 b-lactamase (Table III). In addition, it should be noted that the size of displayed protein complex is limited by the size of outer membrane pIV channel, which mediates phage extrusion. It has been shown that the pIV porin present in the outer membrane of ˚ in diameter at its most constricted E.coli forms a ring 60 A point. The size of this pore sets a size limit for the studied interaction, and it is a potential disadvantage as compared with classical phage display (Opalka et al., 2003). The size constraint was partially mitigated in these experiments by the fusion of b-lactamase to the gene III CT domain rather than the full-sized protein. The removal of two gene III protein domains thereby provided an increased volume for the BLIP-b-lactamase protein complex to be extruded through the pIV porin. The results of the selection from the BLIP Y50X combinatorial library indicate that tyrosine at position 50 in BLIP is not the optimal residue for binding to TEM-1 b-lactamase. This is thought to be due to direct steric clashes with residues on TEM-1, and strong energetic coupling between the neighboring residues on BLIP (Zhang and Palzkill, 2003, 2004; Wang et al., 2009). The removal of the bulky tyrosine side chain (Y50A mutant) stabilizes the BLIP:TEM-1 interaction (Zhang and Palzkill, 2003, 2004; Wang et al., 2009). The enrichment of BLIP Y50M suggests that it has a greater than wild-type affinity for TEM-1. The affinity of Y50M was determined to be 0.12 nM for TEM-1 using an inhibition assay (Table III). In parallel, the same assay was repeated for wild-type BLIP and the Y50A BLIP mutant as reference points. Inhibition constants (Ki) for wild-type BLIP and Y50A were 0.40 and 0.07 nM, respectively. These values are similar to previously determined inhibition constants (Zhang and Palzkill 2003). The mechanism for the increased affinity of BLIP Y50M for TEM-1 b-lactamase relative to wild-type BLIP awaits further biochemical and structural investigations. In conclusion, a general strategy is described for the screening and selection of high-affinity protein – protein interactions based on the binding between a target protein and phage-displayed receptor protein in the periplasm of E.coli followed by recovery of bound phage using an antibody to the soluble, target protein. Validation of the system using the well-characterized BLIP – TEM-1-b-lactamase protein interaction indicates that the method will be useful for the engineering of tight-binding protein – protein interactions. 827
B.G.Fryszczyn et al.
Supplementary data Supplementary data are available at PEDS online. Funding This work was supported by National Institutes of Health grant AI32956 to T.P. and by a training fellowship from the Biomedical Discovery Training Program of the Gulf Coast Consortia (National Institutes of Health grant T90 DA022885-04). References Albeck,S. and Schreiber,G. (1999) Biochemistry, 38, 11– 21. Amann,E., Ochs,B. and Abel,K.J. (1988) Gene, 69, 301– 315. Amann,E., Brosius,J. and Ptashne,M. (1983) Gene, 25, 167–178. Ambler,R.P. (1980) Phil. Trans. R. Soc. London B 289, 321–331. Bradbury,A.R., Sidhu,S., Dubel,S. and McCafferty,J. (2011) Nat. Biotechnol., 29, 245– 254. Brosius,J., Erfle,M. and Storella,J. (1985) J. Biol. Chem., 260, 3539–3541. Brown,N.G. and Palzkill,T. (2010) Protein Eng. Des. Sel., 23, 469–478. Bullock,W.O., Fernandez,J.M. and Short,J.M. (1987) BioTechniques, 5, 376–379. Bush,K. (2010) Curr. Opin. Microbiol., 13, 558– 564. Bush,K. and Jacoby,G.A. (2010) Antimicrob. Agents Chemother., 54, 969–976. de Wildt,R.M.T., Tomlinson,I.M., Ong,J.L. and Holliger,P. (2002) Proc. Natl Acad. Sci. USA, 99, 8530–8535. Doran,J.L., Leskiw,B.K., Aippersbach,S. and Jensen,S.E. (1990) J. Bacteriol., 172, 4909–4918. Duenas,M. and Borrebaeck,C.A. (1994) Biotechnology, 12, 999–1002. Gramatikoff,K., Georgiev,O. and Schaffner,W. (1994) Nucleic Acids Res., 22, 5761– 5762. Huang,W., McKevitt,M. and Palzkill,T. (2000a) Gene, 251, 187 –197. Huang,W., Zhang,Z. and Palzkill,T. (2000b) J. Biol. Chem., 275, 14984–14988. Jung,S., Arndt,K.M., Muller,K.M. and Pluckthun,A. (1999) J. Immunol. Methods, 231, 93– 104. Levin,A.M. and Weiss,G.A. (2006) Mol. Biosyst., 2, 49– 57. Majiduddin,F.K., Materon,I.C. and Palzkill,T.G. (2002) Int. J. Med. Microbiol., 292, 1 –11. Morrison,J.F. (1969) Biochem. Biophys. Acta, 185, 269– 286. Opalka,N., Beckmann,R., Boisset,N., Simon,N.M., Russel,M. and Darst,S. (2003) J. Mol. Biol., 325, 461– 470. Petrenko,V. (2008) Expert Opin. Drug Deliv., 5, 1093– 1099. Petrosino,J., Rudgers,G., Gilbert,H. and Palzkill,T. (1999) J. Biol. Chem., 274, 2394–2400. Reichmann,D., Cohen,M., Abramovich,R., Dym,O., Lim,D., Strynadka,N.C.J. and Schreiber,G. (2007) J. Mol. Biol., 365, 663 –679. Reynolds,K.A., Thomson,J.M., Corbett,K.D., Bethel,C.R., Berger,J.M., Kirsch,J.F., Bonomo,R.A. and Handel,T.M. (2006) J. Biol. Chem., 281, 26745–26753. Rudgers,G.W. and Palzkill,T. (1999) J. Biol. Chem., 247, 6963–6971. Sidhu,S.S. and Koide,S. (2007) Curr. Opin. Struc. Biol., 17, 481–487. Smith,G.P. (1985) Science, 228, 1315–1317. Strynadka,N.C.J., Jensen,S.E., Alzari,P.M. and James,M.N.G. (1996) Nat. Struct. Biol., 3, 290–297. Studier,F.W. and Moffatt,B.A. (1986) J. Mol. Biol., 189, 113– 130. Wang,J., Zhang,Z., Palzkill,T. and Chow,D.-C. (2007) J. Biol. Chem., 282, 17676–17684. Wang,J., Palzkill,T. and Chow,D.C. (2009) J. Biol. Chem., 284, 595– 609. Yuan,J., Huang,W., Chow,D.C. and Palzkill,T. (2009) J. Mol. Biol., 389, 401–412. Zhang,Z. and Palzkill,T. (2003) J. Biol. Chem., 278, 45706–45712. Zhang,Z. and Palzkill,T. (2004) J. Biol. Chem., 279, 42860–42866.