doi:10.1016/j.jmb.2003.09.033

J. Mol. Biol. (2003) 334, 241–254

High-density Functional Display of Proteins on Bacteriophage Lambda Amita Gupta1, Masanori Onda2, Ira Pastan2, Sankar Adhya2 and Vijay K. Chaudhary1* 1 Department of Biochemistry University of Delhi South Campus, New Delhi 110021 India 2

Laboratory of Molecular Biology, Center for Cancer Research, 37 Convent Dr., Rm. 5106, Bethesda, MD 20892-4264, USA

We designed a bacteriophage lambda system to display peptides and proteins fused at the C terminus of the head protein gpD of phage lambda. DNA encoding the foreign peptide/protein was first inserted at the 30 end of a DNA segment encoding gpD under the control of the lac promoter in a plasmid vector (donor plasmid), which also carried lox Pwt and lox P511 recombination sequences. Cre-expressing cells were transformed with this plasmid and subsequently infected with a recipient lambda phage that carried a stuffer DNA segment flanked by lox Pwt and lox P511 sites. Recombination occurred in vivo at the lox sites and Ampr cointegrates were formed. The cointegrates produced recombinant phage that displayed foreign protein fused at the C terminus of gpD. The system was optimised by cloning DNA encoding different length fragments of HIV-1 p24 (amino acid residues 1– 72, 1– 156 and 1 – 231) and the display was compared with that obtained with M13 phage. The display on lambda phage was at least 100-fold higher than on M13 phage for all the fragments with no degradation of displayed products. The high-density display on lambda phage was superior to that on M13 phage and resulted in selective enrichment of epitope-bearing clones from gene-fragment libraries. Single-chain antibodies were displayed in functional form on phage lambda, strongly suggesting that correct disulphide bond formation takes place during display. This lambda phage display system, which avoids direct cloning into lambda DNA and in vitro packaging, achieved cloning efficiencies comparable to those obtained with any plasmid system. The high-density display of foreign proteins on bacteriophage lambda should be extremely useful in studying low-affinity protein –protein interactions more efficiently compared to the M13 phage-based system. q 2003 Elsevier Ltd. All rights reserved.

*Corresponding author

Keywords: phage display; recombination; gpD; gene-fragment library; antibody fragment

Introduction Display of peptides and proteins on the surface of bacteriophage is a powerful approach to study protein– protein interactions.1 – 3 The power of the system emanates from the direct physical linkage Abbreviations used: cfu, colony-forming units; DCO, double crossover; HRP, horse radish peroxidase; mAb, monoclonal antibody; PAG, polyacrylamide gel; PE, Pseudomonas exotoxin; pfu, plaque-forming units; SCO, single crossover. E-mail address of the corresponding author: [email protected]

between the phenotype and its genotype, so that a desired molecule can be easily selected from a milieu of millions and identified by sequencing the DNA encapsulated in that phage particle. Phage display technology, which started with the identification of peptide epitopes recognised by monoclonal antibodies4 has grown into an approach for cloning human antibodies, studying ligand – receptor interactions, elucidating signal transduction pathways, delineating contact residues in interacting proteins, and isolating peptide inhibitors.2,5,6 Other applications of this technology include the production of gene/genome-fragment and cDNA libraries7 – 10 displaying virtually every

0022-2836/$ - see front matter q 2003 Elsevier Ltd. All rights reserved.

242

Functional Display of Proteins on Lambda

Figure 1. Diagrammatic representation of different vectors. Only relevant genes and restriction sites are shown. The maps are not to scale. lacPO, lac promoter-operator; RBS, ribosome-binding site; D, segment encoding amino acid residues 1 – 109 of gpD of lambda; Stuffer, a 30 nucleotide long sequence; c-myc, decapeptide recognised by monoclonal antibody, 9E10; fori, origin of replication of filamentous phage f; Ampr, b-lactamase gene; Ori, ColE1 origin of replication; lox Pwt, wild-type lox site; lox P511, lox site with mutation 511. A, Donor plasmid, pVCDcDL1 with cloning sites NheI and MluI. Details of regions marked as a and b by double-headed arrows are shown in Aa and Ab, respectively. Amino acids are shown in single-letter code below the nucleotide sequence. Restriction enzyme sites are shown above the nucleotide sequence. p, translation stop. L7 and L15 are oligonucleotide primers used for PCR-based analysis of recombinants. Aa, Sequence showing the ribosome-binding site and the initiation codon in pVCDcDL1. Ab, Sequence showing the end of gpD coding region, collagenase site, stuffer fragment and c-myc in pVCDcDL1. B, Recipient phage vector, lDL1. Only some of the lambda genes are shown. Dam, D gene of lambda with amber mutation. The unique XbaI site in the lambda genome used for cloning is shown. lacZa, DNA cassette comprised of lacPO, RBS and the first 58 codons of lacZ; L1 and L4 are oligonucleotide primers used for PCR-based analysis of cointegrates. C, Donor plasmid pVCDcDL3 is similar to pVCDcDL1 but contains between the NheI and MluI sites a lacZ cassette comprised of lacPO, RBS and the first 148 codons of lacZ flanked by SmaI/SrfI restriction enzyme sites. Blunt-ended DNA fragments can be cloned into SmaI/SrfI-cut vector and recombinants produce white colonies on X-gal plates. T, universal translation stop.

possible encoded peptide/protein that can be used for identifying specific interacting sequences. The bacteriophage M13 has been the most widely used system for display of peptides/proteins. Fusion to the minor coat protein, gIIIp, and the major coat protein, gVIIIp, of M13 is used for display of a range of molecules of different sizes and structures. However, the high-density display of large protein domains on M13 is inefficient and often associated with extensive degradation.11 Since M13 morphogenesis occurs in the periplasm, it is possible that the molecules that are secretionincompetent may not get displayed. M13 is used

primarily as an N-terminal display system but some variants have been developed for C-terminal display12,13 and construction of cDNA libraries.14 There have been reports in recent years describing N and C-terminal display of peptides/proteins on phage lambda as fusion to the capsid protein “d” (gpD) and tail protein “v” (gpV) of lambda.15 – 17 These systems have yet to gain wide acceptance for the following reasons: (i) lambda phage biology is more complex than that of M13; (ii) the lambda genome is very large (50 kb); as a result, isolation of viral DNA, insertion of user-defined restriction sites, cloning of foreign fragments and then

Functional Display of Proteins on Lambda

packaging of the ligated product in vitro to make lambda particles are difficult and the library sizes achieved are smaller than those obtained with phagemid-based M13 vectors;18 and (iii) intracellular assembly of phage may not allow disulphide bond formation in the molecule to be displayed. Here, we describe a lambda phage display system that allows simplified, high-efficiency cloning in lambda DNA and high-density display of peptides/proteins fused to gpD of lambda. We compare the display of different molecules on lambda phage and M13 phage and demonstrate the advantage of high-density display on lambda in biopanning for epitope mapping using genefragment libraries. We also show that the lambda phage system is able to display proteins with multiple disulphide bonds in functional form and in large numbers.

Results Cloning into lambda display vector by in vivo recombination The cloning strategy is based on first inserting DNA encoding peptide-protein into a high copy donor plasmid vector, pVCDcDL1 (Figure 1A), and then transferring this genetic information into recipient lambda genome, lDL1 (Figure 1B), by the high-efficiency lox-Cre recombination system in vivo (Figure 2).

243

pVCDcDL1 contained a sequence encoding gpD of l (Figure 1Aa), followed by a GGSG spacer(s), a collagenase site, NheI site, a stuffer segment, MluI site and a c-myc tag (Figure 1Ab), under the control of the lac promoter (lacPO). Cloning of DNA sequences as NheI– MluI inserts in place of the stuffer allowed formation of a D fusion protein with collagenase site between D and the foreign protein and c-myc tag at the C terminus. The vector also contained the M13 phage origin of replication ( fori), flanked by lox Pwt and lox P511 recombination sequences. lDL1, the recipient lambda vector, contains a lacZa fragment flanked by lox Pwt and lox P511 recombination sequences at the unique XbaI site present in the lambda genome. The lox sequences in the donor plasmid are in the reverse orientation to that in the recipient lambda genome (Figure 1B). When Escherichia coli expressing Cre recombinase (Creþ host) were transformed with the donor plasmid and then infected with lDL1, recombination occurred at the compatible lox sites in the two vectors, resulting in integration of the plasmid DNA into the lambda DNA (Figure 2). Note that Cre-mediated recombination occurs between two lox Pwt sites or lox P511 sites, and not between lox Pwt and lox P511 sites.19 Hence, plasmid and lambda DNA crossing over occurred only in trans and resulted in formation of a cointegrate. Also, due to opposite orientation of the lox sites in the plasmid and lambda, the recombination led to integration of the entire plasmid DNA into the lambda DNA. The first

Figure 2. The process of recombination. lox sites shown in black are of the recipient lambda phage vector. Cre, Cre recombinase; SCO, single crossover cointegrate; DCO, double crossover cointegrate. Filled arrows indicate the direction of transcription from the promoter of b-lactamase, lacZa and D gene. L1 and L4 are oligonucleotides used for PCR-based analysis of cointegrate. Only one of the possible recombination pathways is shown (first crossover at lox Pwt followed by second crossover at lox P511). The other pathway (first crossover at lox P511 followed by second crossover at lox Pwt) will yield the same product.

244

Functional Display of Proteins on Lambda

Table 1. Analysis of recombination of recipient vector lDL1 and donor plasmid pVCDcDL1

Table 2. Characterisation of phage produced by SCO and DCO cointegrates

Sample

pfu/ml

cfu/ml

Phage

Titre (pfu)a

EDTA resistantb

Recovery (%)c

Lysate obtained after recombination in Cre2 host Lysate obtained after recombination in Creþ host

1.5 £ 109

0

1.5 £ 109

2.5 £ 108

lDcDL1: SCO lDcDL1: DCO lDL1

4 £ 109/ml 4 £ 109/ml 4 £ 109/ml

þ þ þ

2.24 2.22 ,0.01

Cultures of TG1 (Cre2) and BM25.8 (Creþ) transformed with pVCDcDL1 were infected with lDL1 and the lysate obtained titrated as pfu and cfu in TG1.

crossover event (intermolecular) resulted in the formation of single crossover (SCO) cointegrate that contained the complete donor plasmid integrated in the lambda genome. A second crossover event (intramolecular) at the other pair of compatible lox sites resulted in the formation of a double crossover (DCO) cointegrate and excision of the lacZa fragment and fori sequence (Figure 2). Thus, the DNA encoding the foreign peptide/ protein fused to gpD for display on lambda phage surface becomes part of the lambda genome. The lambda also acquires the b-lactamase selection marker of the plasmid. Based upon this strategy, BM25.8 (Creþ host) and TG1 (Cre2 host) were transformed with the donor plasmid, pVCDcDL1, and then infected with recipient lambda phage, lDL1. The cultures were grown in ampicillin-containing medium until complete cell lysis. The cell-free lysate obtained after the recombination event was used to infect Cre2 cells and plated to determine plaque-forming units (pfu) and colony-forming units (cfu) (on ampicillin-containing medium). As shown in Table 1, the number of pfu was the same in the lysate obtained from Creþ and Cre2 hosts, indicating similar amounts of phage production in both hosts. However, only the lysate from Creþ host was able to transduce Ampr colonies in E. coli. This indicates that the plasmid integrated into lambda DNA only in the presence of Cre protein and conferred ampicillin resistance to cells harbouring this lambda cointegrate as extra chromosomal lysogen driven by the plasmid replicon. The lysate obtained from Creþ host contained three phage species: parental recipient lambda, SCO cointegrate (lDcDL1: SCO) and DCO cointegrate (lDcDL1: DCO). Plating on ampicillincontaining medium selected for cointegrates and eliminated parental phage. To check for the presence of plasmid sequence in lambda genome, the Ampr colonies were analysed by PCR using primers L1 and L4 (Figure 2) that flank the lox sequences in lambda. Agarose gel electrophoresis of amplified products revealed that all the colonies analysed harboured cointegrates and the ratio of SCO to DCO cointegrates was 1:3 (data not shown). lDcDL1: SCO and lDcDL1: DCO harbouring clones were grown in ampicillin-containing medium wherein there was spontaneous phage production leading to cell lysis. The cell-free

lDcDL1: SCO, single crossover cointegrates of lDcDL1 displaying gpD-c-myc; lDcDL1:DCO, double crossover cointegrate of lDcDL1 displaying gpD-c-myc; lDL1, wild-type recipient phage not displaying c-myc. a Colonies harbouring SCO and DCO cointegrates of lDcDL1 were grown and phage titre in culture lysate determined as pfu. b Lambda phage were diluted in TM (10 mM Tris, 10 mM MgCl2) or TE (10 mM Tris, 10 mM EDTA) and incubated at 37 8C for ten minutes. Serial dilutions of the samples were then titred on TG1 cells to determine pfu. þ Indicates no difference in the titre of phage in TM and TE diluted samples. c Bio-panning was done using anti-c-myc mAb, 9E10. The recovery was calculated as (output titre/input titre) £ 100.

lysates obtained were tested for phage titre and presence of gpD-c-myc protein on the phage surface. Both SCO and DCO harbouring cells produced the same number of phage (Table 2) determined as pfu. To test the stability of phage particles, the lysates were incubated in EDTAcontaining buffer and then re-titrated to determine the number of viable phages. No difference in pfu before and after incubation in EDTA was observed for lysates obtained from SCO and DCO clones, indicating that the phages produced were resistant to EDTA and all 405 copies of gpD (either as gpD or gpD fusion protein) were present on every phage particle.20 The phage particles were then tested for display of c-myc peptide as gpD fusion. Both types of phage displayed the same amount of c-myc peptide as revealed by equal recovery of phages (, 2% of phages added) following biopanning in anti-c-myc (mAb 9E10) coated wells (Table 2). This recovery was at least 200-fold higher than that obtained for lDL1 phage (that does not display gpD-c-myc). Western blot analysis with mAb 9E10 showed that phages purified from lysate of both SCO and DCO clones produced a band of , 16 kDa with an intensity corresponding to , 400 copies of fusion protein per phage particle (data not shown; the number of fusion proteins was calculated by densitometric scanning of the blot using a purified c-myc-containing protein as control). These experiments established that SCO and DCO phage had similar properties and all ampr transductants obtained after recombination produced functional phage displaying gpD fusion protein. High-density display of peptides and proteins on lambda; a comparison with M13 Display of different size molecules on lambda phage and a comparison with the M13 phage display system in terms of density and functionality of displayed peptides and proteins was carried

Functional Display of Proteins on Lambda

245

Figure 3. Analysis of phage displaying fragments of HIV-1 capsid protein, p24. (A) ELISA: microtitre plates were coated with ascitic fluid of anti-p24 mAb, H23 (1:1000 dilution) and assay performed as described in Materials and Methods. Phage displaying p241 (amino acid residues 1 – 72 of p24) as fusion to gVIIIp (A), gIIIp (B) and gpD (F); phage displaying p246 (amino acid residues 1 – 156 of p24) as fusion to gVIIIp (W), gIIIp (X) and gpD (%); phage displaying p24 (amino acid residues 1 – 231 of p24) as fusion to gVIIIp (K), gIIIp (O) and gpD (i). (B) Western blot: 1 £ 108 lambda phage and 1 £ 1011 M13 phage were electrophoresed on 0.1% SDS-12.5% PAG (a and b) or 2 10% PAG (c), transferred onto PVDF membrane and probed with anti-c-myc mAb, 9E10, followed by HRP-conjugated goat anti-mouse IgG. a, Lambda phage displaying fusion protein with gpD; b, M13 phage displaying fusion protein with gVIIIp; c, M13 phage displaying fusion protein with gIIIp. gIIIp has aberrant mobility (55 – 60 kDa instead of 45 kDa) accounting for the difference in the calculated and the observed molecular mass. Lane C, phage displaying gpD-cmyc fusion protein; lane 1, phage displaying p241 fused to coat protein; lane 2, phage displaying p246 fused to coat protein; lane 3, phage displaying p24 fused to coat protein. MW, molecular mass markers with molecular mass in kDa. Arrowhead denotes the degradation product. (C) Quantification of displayed fusion protein on lambda and M13 phage. The number of fusion protein molecules per phage particle was quantified by densitometric scanning of Western blot (developed with anti-c-myc mAb, 9E10, in B using purified fusion protein GST-c-myc as calibration standard. MW, calculated molecular mass of fusion protein in kDa. nd, not determined.

out using fragments of HIV-1 capsid protein p24. HIV-1 p24 contains two independently folding domains.21,22 The first 156 amino acid residues of p24 constitute the N-terminal domain that interacts with host proteins such as cyclophilin, while residues 157– 231 constitute the C-terminal domain, which is responsible for oligomerisation of p24 to form the viral capsid. Three fragments of p24 encompassing residues 1 –72 (p241), 1 –156 (p246, N-terminal domain of p24) and 1 –231 (p24, fulllength protein) were displayed as C-terminal fusions with gpD on lambda, and as N-terminal fusions with gVIIIp and gIIIp on M13 using phagemid-based vectors. All the fusion proteins contained c-myc tag at the C terminus of p24 fragment. Phage were prepared for all lambda and M13 clones and purified by polyethylene glycol (PEG) precipitation and ultracentrifugation. The purified phages were then tested for binding to anti-p24 mAb in ELISA and the display of fusion protein on the phage surface was quantified by Western blot using anti-c-myc mAb 9E10. In ELISA, both M13 and lambda phage displaying p24 fragments showed dose-dependent binding to mAb H23, which recognises amino acid residues 56– 66 of p24 (Figure 3A). p241-displaying phage showed maximum reactivity followed by p246-

displaying and p24-displaying phage. As seen in Figure 3A, for all the three displayed molecules, lambda phage showed two to three orders of magnitude better reactivity compared to corresponding M13 phage, indicating higher display of the proteins. The number of fusion protein molecules displayed per phage particle was quantified by Western blot analysis using mAb 9E10 (Figure 3B). In the case of lambda phage (Figure 3Ba), an intense band corresponding to the calculated molecular mass was seen for each of the three fusion proteins. The number of fusion protein molecules displayed per phage particle was estimated to be 350 copies of gpD-p241-c-myc (22 kDa, Figure 3Ba, lane 1) followed by 210 copies of gpD-p246-c-myc (31 kDa, Figure 3Ba, lane 2) and 154 copies of gpDp24-c-myc (39 kDa, Figure 3Ba, lane 3 and Figure 3C). In the case of M13, the lane corresponding to phage displaying p241 (lane 1, Figure 3Bb and Bc) showed only one band having molecular mass (,13 kDa as gVIIIp fusion and ,60 kDa as gIIIp fusion) less than calculated for the fusion protein (Figure 3C). Since full-length fusion protein was not visible on the blot, the amount of p241 fusion protein on M13 phage could not be determined. The lane corresponding to M13 phage displaying

246

p246 and p24 showed two major bands in each blot (Figure 3Bb and Bc). The band with slower mobility corresponded to the calculated molecular mass of the fusion protein but the second, more intense band, showed mobility similar to that seen in the lane with M13 phage displaying p241, suggesting these to be degradation products that had retained the c-myc epitope. This faster moving band was found to be reactive to mAb 9E10 but not to mAb H23, confirming the loss of amino acids from the N terminus (data for mAb H23 not shown). Densitometric scanning showed that M13 phage displayed less than two copies of the fusion protein per phage particle (Figure 3C). The Western blot data obtained with mAb 9E10 correlated well with the ELISA data obtained for reactivity of phage to mAb H23. The full-length p241-gVIIIp/gIIIp fusion protein may be present in extremely low quantities on M13 phage (not detected in Western blot); however, the degradation product that was displayed on the phage surface retained H23 epitope (confirmed by Western blot of phages using H23; data not shown) resulting in the high reactivity observed in ELISA. This analysis clearly shows that the lambda phage system is capable of displaying proteins of different sizes with large domains in much higher density than the M13 phage system, with less degradation of the fusion protein. High-density display leads to efficient selection in bio-panning The utility of high-density display on lambda was evaluated in epitope mapping of monoclonal antibodies, and polyclonal serum from human subjects. For this purpose, a gene-fragment library of Pseudomonas exotoxin A (PE) derived peptides as fusion to gpD of lambda or to the gIIIp of M13 was constructed. For making the lambda library, DNA encoding PE-38, a 38 kDa fragment of PE,23 was digested with DNase I to produce 50– 200 bp random fragments that were flushed and ligated to SmaI (CCCGGG)-digested donor plasmid vector, pVCDcDL3 (Figure 1C). BM25.8 (Creþ) cells were transformed with the ligation sample

Functional Display of Proteins on Lambda

and a library efficiency of 1 £ 107 transformants per microgram of vector DNA was obtained. Analysis of individual transformants showed that 98% clones were recombinants and contained PEderived inserts that were randomly distributed along the PE sequence (data not shown). This plasmid library was then infected with lDL1 to allow recombination and formation of cointegrates. The lambda library obtained consisted of phages displaying 20– 70 amino acid residue peptides fused to the C terminus of gpD and a similar M13 library (made using the protocol as described in Materials and Methods) comprised of phages carrying PE-derived peptides fused to the N terminus of gIIIp. Analysis of individual tranductants of the phage library confirmed that the inserts were faithfully transferred from the plasmid to the lambda genome during recombination and their distribution was random along the PE sequence. Also, the lambda library contained 18% of clones that had the PE fragment in-frame with the gpD coding sequence and therefore displayed PE-derived peptide on the phage surface as against only 5% of clones of a similar kind in the M13 library. This is because C-terminal fusion to gpD in lambda allows three times more PE-derived sequences to be inframe with the gpD sequence and get displayed as compared to N-terminal fusion to gIIIp in M13. The lambda and M13 libraries were first used to map the epitope recognised by a mAb against PE. The binding of lambda phage particles to mAbcoated wells was 2000 times more than the binding to uncoated wells, while with the M13 library this ratio was only ten times (Table 3). Sequence analysis of individual phage clones obtained in eluates from mAb-coated wells revealed that 88% of the analysed lambda clones contained the PE fragment inserted in-frame with the gpD sequence, as compared to 63% of M13 clones. This difference reflects more specific binding of lambda phages to mAbcoated wells. These individual phage clones were then tested for binding to anti-PE mAb in ELISA. The lambda clones produced high dose-dependent reactivity in ELISA to anti-PE mAb, while none of the M13 clones showed any significant reactivity despite the addition of 1000-fold more M13 phage

Table 3. Panning of lambda and M13 phage PE gene-fragment library on anti-PE mAb Lambda library Test Input phage Output phage Fold enrichmenta % Clones with PE fragment in frame with coat protein (no. positive/no. analysed)b % Clones ELISA reactive (no. positive/no. analysed)c

M13 library

Control 8

8

1 £ 10 5

1 £ 10 1 £ 104

Test

Control 10

1 £ 1010 9 £ 104

1 £ 10 9 £ 105

2 £ 103 88 (22/25)

10 63 (23/36)

100 (12/12)

0 (0/14)

Test, coating with anti-PE mAb; Control, no coating. a Fold enrichment ¼ output phage in test/output phage in control. b The inserts in clones from test wells selected after panning were sequenced to check for reading frame of PE fragment with respect to the coat protein. c The clones displaying PE peptides were analysed for binding to anti-PE mAb in ELISA.

247

Functional Display of Proteins on Lambda

(data not shown). The low density of fusion protein on M13 phage surface in combination with the low affinity of anti-PE mAb might have resulted in very low capture of M13 phages on mAb-coated wells, which could not be detected in ELISA. These results clearly demonstrate that the highdensity display on lambda phage leads to more specific binding, which results in higher enrichment during panning and higher ELISA reactivity of enriched clones. The libraries were then used to map immunodominant epitopes of PE using polyclonal serum. Appearance of anti-toxin antibodies in serum of patients undergoing immunotoxin therapy is a major obstacle in the way of repeated treatment with immunotoxin and success of immunotoxins in cancer therapy. Identification of immunodominant epitopes in toxin molecules will enable engineering of an altered toxin, which retains full bio-efficacy but will not elicit neutralising antibodies when administered to patients. For this purpose, the PE gene-fragment library was used to identify regions of PE against which antibodies are present in the serum of human patients after treatment with recombinant immunotoxins containing PE-38.24 The anti-PE response in these patients is characterised by low titre of antibodies that have weak affinity, since the patients are administered few doses of immunotoxin in a short period of time. Microtitre wells were coated with pre-treatment (control) and post-treatment (test) sera from patients administered immunotoxin therapy, and panning was performed on individual serum samples. The results obtained were similar for all samples. As shown in Table 4, for one sample, the enrichment obtained was more for the M13 library (123-fold) as compared to lambda library (20-fold). However, only 16% of the M13 clones obtained from the test sample had the PE fragment in-frame with gIIIp while 90% of the selected lambda clones had the PE fragment in-frame with gpD. The DNA sequence of individual phage clones recovered from test wells was translated and aligned to the PE sequence. The alignment data showed that all the l clones aligned to the last 50 amino acid residues of PE (Figure 4A), clearly indicating this region as the epitope. In the case of M13, the clones

Figure 4. Epitope mapping using anti-PE serum. (A) Alignment of inserts in phage selected after panning of PE gene-fragment library in M13 and lambda phages on serum from PE-immunotoxin-treated patients. Only 15 clones (out of 30 clones) of lambda are shown. The continuous line represents PE-38 and the numbers above the line denote amino acid position in the full-length PE. All the clones were tested in ELISA but the clones whose data are shown in (B) are indicated as a to d. (B) Reactivity of phage displaying PE fragments in ELISA. The details are described in Materials and Methods. Reactivity of M13 phage, (a) (A,B) and (b) (W,X) and lambda phage, (c) (K,O) and (d) (L,P) to pre-treatment sera (open symbols) and post-treatment sera (filled symbols).

aligned to different regions in PE and no common alignment could be deduced. A few clones aligned to the last 50 residues of PE (Figure 4A) but their number was not sufficient for this region to be

Table 4. Panning of lambda and M13 phage PE gene-fragment library on human serum Lambda library Control Input phage Output phage Fold enrichmenta Clones with PE fragment in frame with coat protein % (no. positive/no. analysed)b

M13 library

Test

8

Control 8

1 £ 10 1 £ 103

1 £ 10 2 £ 104

1 £ 10 6 £ 104

90 (30/33)

9 (3/34)

20 25 (8/30)

Test

10

1 £ 1010 7 £ 106 123 15.7 (11/70)

Control and Test refer to wells coated with pre-treatment serum and post-treatment serum of patients, respectively. a Fold enrichment ¼ output phage in test/output phage in control. b The inserts in clones from test wells selected after panning were sequenced to check for reading frame of PE fragment with respect to the coat protein.

248

demarcated as an immunodominant epitope. A second round of panning with the amplified eluate of the first panning of M13 phage did not result in any increase in the number of clones with alignment in any one region of PE (data not shown). Therefore, the exact epitope(s) against which antibodies are present in serum could not be deduced from M13 library data. This difference in the results for the two libraries could be due to low ligand concentration per phage particle in the case of M13 as well as the presence of a high number of non-displaying clones in the total M13 population, which resulted in more non-specific binding and low binding of specific phage in the well. Samples from several patients were used for panning and similar results obtained (data not shown). ELISA was performed to test the clones for reactivity to serum antibodies. In ELISA, the lambda clones showed high dose-dependent reactivity with post-treatment serum pool and gave low reactivity with pre-treatment serum pool (Figure 4B; data for two representative clones are shown). On the other hand, M13 phage clones showed reactivity with pre-treatment serum with only marginal increase in reactivity with posttreatment serum, indicating non-specific binding of M13 phage to human immunoglobulins coated in the microtitre well. The difference in ELISA reactivity between lambda and M13 phages was two to three orders of magnitude, again establishing the importance of high-density display on lambda phage in the identification of immunodominant epitopes using polyclonal serum. The biopanning data described above clearly show that high-density display on lambda phage increases panning efficiency from a library and offers a distinct advantage over M13 phage in selection of specific binders.

Functional Display of Proteins on Lambda

Figure 5. ELISA of lambda phage displaying SS1scFv. Microtitre plates were coated with anti-c-myc mAb 9E10 (A), or recombinant mesothelin (B). Purified l DcSS1DL1 phage (X), l DcDL1 phage (K) and lDL1 phage (A) were added and ELISA performed as described in Materials and Methods.

Display of disulphide bond-containing proteins One major application of phage display technology is identification of protein –protein interaction cascades wherein a plethora of protein sequences are displayed on the phage surface, several of which might contain disulphide bonds essential for their function. We used the singlechain fragment (scFv) of an antibody as fusion partner with gpD to test the display of disulphidecontaining proteins in functional form on lambda. An scFv molecule contains two intra-molecular disulphide bonds, which are essential for its correct conformation and activity. Therefore, functional display of scFv as gpD fusion on lambda surface will indicate that disulphide bonds are formed in proteins displayed on lambda. Mesothelin is a glycoprotein present on the surface of cancer cells and is a promising candidate for targeted therapies.25 SS126 is a high-affinity variant of anti-mesothelin antibody SS.25 Lambda phage displaying SS1 scFv (lDcSS1DL1) were produced by recombination as described in Materials and Methods and purified. These phages display

SS1 scFv fused at the C terminus of gpD with a c-myc tag at the C terminus of scFv. In ELISA on anti-c-myc-coated plates, the binding of lDcSS1DL1 was about 30 times less than that of lDcDL1 (Figure 5A). Thus, lDcSS1DL1 displayed about 10 –15 copies of D-scFv-c-myc fusion protein in comparison to lDcDL1 that displayed 400 copies of D-c-myc fusion protein per phage particle. Functionality of SS1scFv displayed on lambda was checked by binding of phage to the natural ligand of SS1, mesothelin. lDcSS1DL1 phage were added to mesothelin-coated wells and captured phage detected using anti-lambda phage polyclonal sera. lDcSS1DL1 phage showed specific dose-dependent binding to mesothelin, indicating that the displayed scFv molecules were functional (Figure 5B). lDL1 and lDcDL1 phages that did not display SS1 scFv showed no binding to mesothelin. Further, 5 £ 109 lDcSS1DL1 phages gave the same binding to mesothelin as 1 £ 1011 M13 phage displaying SS1 scFv fused to gIIIp

Functional Display of Proteins on Lambda

(data not shown), indicating that the number of functional scFv molecules present per lambda particle was several-fold more than per M13 particle. This was confirmed by Western blot analysis using anti-c-myc mAb 9E10. Here, 1 £ 109 lDcSS1DL1 phage showed a band corresponding to gpD-scFv-c-myc fusion protein while a same intensity band of scFv-c-myc-gIIIp was seen with 5 £ 1010 M13 phage displaying SS1scFv fused to gIIIp (data not shown). This result establishes that disulphide bond-containing proteins are also displayed in higher numbers on lambda phage as compared to M13 phage.

Discussion Here, we describe a user-friendly lambda-based phage display system with a simple strategy for high-efficiency cloning of foreign DNA into lambda phage genome and display of the encoded peptide/protein in functional form in high numbers on the surface of lambda phage fused to the C terminus of the D protein (gpD). In this system, the display of peptides, large protein domains and full-length proteins, including disulphidebonded proteins, on the surface of bacteriophage lambda was several orders of magnitude higher than on the surface of bacteriophage M13 (which to date is the most widely used display vehicle). Also, this high-density display on lambda phage was extremely useful in mapping epitopes of monoclonal antibodies and in identifying immunodominant epitopes of a toxin molecule using a toxin gene-fragment library displayed on lambda. A prerequisite for any surface display system to be used in the form of a library is efficient cloning of DNA sequences encoding the peptides/proteins to be displayed, into the genome of the display vehicle. With a large size of 50 kb, it is not easy to achieve cloning efficiencies in lambda as high as those obtained with plasmid DNA using current transformation protocols. As a result, making large libraries in lambda is an arduous task. The lambda phage system described here overcomes this limitation because the cloning of the foreign DNA sequence (encoding the protein/peptide to be displayed) is carried out in a plasmid vector. Then, by a highly efficient process of phage infection and in vivo recombination, the cloned sequence is integrated into the lambda genome and the peptide encoded by the cloned sequence is displayed as gpD fusion protein on the surface of progeny lambda phage particles. Because the cloning is done in a plasmid, high transformation efficiencies (by electroporation) can be achieved. Also, the recombination occurs in vivo, eliminating the need to isolate lambda DNA, clone DNA sequences into it and package the recombinant lambda in vitro. The frequency of recombinants (with insert) at the plasmid level as well as the phage level (cointegrates) is greater than 90% as

249

against 3 – 15% reported for direct cloning in lambda display vectors.27 We have constructed a gene-fragment library of the M. tuberculosis genome in lambda with library size of 3 £ 107 clones per microgram of vector having 90% recombinants (unpublished data). The recombination in vivo is mediated by lox recombination sequences present in the donor plasmid and the recipient lambda vector. We have used two non-compatible lox recombination sequences, lox Pwt and lox P511;19 recombination can thus occur only in trans resulting in integration of plasmid sequence into lambda DNA. This strategy circumvents the problem of excision of integrated plasmid as observed earlier9 due to the presence of two compatible lox Pwt sites in the cointegrate. The recombinants (cointegrates) formed can be easily selected for antibiotic resistance conferred by the integrated plasmid. The cointegrates contain the cloned foreign DNA sequence as part of their genome and the corresponding phage display the encoded peptide/protein as a gpD fusion protein on their surface. In the system described here, more than 75% of cointegrates were DCO cointegrates and are inert to any recombination in cis. The single crossover cointegrates have two lox Pwt and two lox P511 sites and can undergo recombination in a Cre2 host as observed earlier by other workers,9 which can result in either formation of a DCO or loss of integrated plasmid. It should be possible to eliminate SCO by using a recipient lambda vector having a counter-selectable marker in place of lacZa and selecting for cells containing DCO co-integrates on non-permissive growth medium. Another strategy could involve increasing the length of the sequence (which is excised out during formation of DCO) between the lox sites in the recipient lambda vector so that, after recombination, in vivo packaging of SCO with genome size greater than 51 kb would become inefficient. The lambda system described here was able to display proteins of different sizes (72, 156 and 231 amino acid residues), and the number of copies of each protein per phage particle on lambda was two to three orders of magnitude greater as compared to display on M13 phages as fusion to gVIIIp or gIIIp. The 156 and 231 amino acid residue fragments of p24 represent complete functional domains of the HIV-1 capsid protein, p24. With such a high density of fusion protein on the surface, the lambda phage particle can be envisaged as a dense mass of functional protein, with the encoding DNA encapsulated inside, that can be directly used in affinity selection of specific clones from a large library or for studying low-affinity protein –protein interactions. Due to low-density display on M13, the probability of isolating low-affinity interacting partners is greatly reduced. This could be compensated by the avidity effect due to high-density display on lambda, and thereby the chances of recovering low-affinity partners in addition to the high-affinity ones would increase. p24 oligomerises during HIV

250

capsid formation to form multimers with solution Kd of 1.3 £ 1025 M.28 In an assay to test binding of p24-displaying phages to recombinant p24,29 lambda phage displaying intact p24 showed approximately 100-fold better binding to immobilised p24 than corresponding M13 phage (unpublished results), suggesting that the lambda-based system would be useful in studying low-affinity interactions. Moreover, gpD not only enables display of few hundred copies of the foreign protein per phage particle both as N and C-terminal fusion,16,17 but is also expressed at high levels and is present in soluble form in the cell cytosol. gpD has been shown to have chaperone-like properties and is a good fusion partner for expression of proteins in soluble form.30 This property of gpD might be responsible for the display of foreign proteins in large numbers on the lambda surface and would also facilitate purification of fusion protein sequences selected from the lambda display libraries. The high-density display on lambda can be particularly useful in epitope mapping. Specific epitopes can be identified from complex genefragment libraries made from whole genomes of infectious viruses and bacteria using immunoglobulins in sera from infected persons as bait. The performance of lambda display in epitope mapping was evaluated using a gene-fragment library of PE and panning on monoclonal antibodies and serum isolated from patients treated with PE-based immunotoxins. Here again, the lambda library resulted in specific enrichment and the selected phages showed higher ELISA reactivity than M13, establishing the superiority of the lambda display system. The lambda library was also used for epitope mapping of a panel of anti-PE mAbs wherein not only linear epitopes (also determined by a synthetic 20-mer peptide library of PE) but also larger conformational epitopes were identified. These conformational epitopes could not be determined by the synthetic peptide library of PE (data to be published elsewhere). This highlights the importance of phage display-based gene-fragment libraries in epitope mapping. With patient’s serum, the M13 system failed to identify any specific epitope. This can be attributed to low titre and low affinity of anti-toxin antibodies present in these serum samples, which were not able to bind M13 phages displaying fewer epitope molecules as compared to corresponding lambda phage. This high selectivity and sensitivity of the lambda display system would be very useful in rapid identification of immunodominant epitopes in acute viral/bacterial infections where response in patients may not comprise high-affinity antibodies with high titres. Other groups have also reported more efficient selection of antibody epitopes using a lambda-based random peptide library in comparison to the M13-based library,8,31 attributing this to the under-representation of secretion-incompetent clones in the M13 library. Our data suggest that it may be the large dif-

Functional Display of Proteins on Lambda

ferences in display density on lambda and M13 that dictate selection efficiencies. Correct disulphide bond formation can be essential for functional display of large protein domains in several cases but may not occur in the reducing environment of the cytosol where lambda assembly takes place. The display of the singlechain Fv fragment (scFv) in functional form on lambda demonstrates that correct disulphide bond formation takes place during cytosolic assembly of lambda particles. It would be difficult to comment on the mechanism of disulphide bond formation; however, cell lysis accompanying lambda phage production might be responsible for providing oxidising conditions that enable disulphide bond formation in the molecules displayed on the phage surface. The display of these proteins may be further improved by using oxidising strains of E. coli that allow disulphide bond formation in the host cytosol32 in conjunction with co-expression of disulphide bond-promoting chaperones. Although M13 is the system of choice for display of antibody fragments and their affinity maturation in vitro, functional display of antibodies on lambda in much higher numbers than on M13 will be particularly useful in isolating low-affinity antibodies, identifying new tumour markers by the ability to pan with higher sensitivity on cell surfaces when the concentration of panning antigen is limited, and in targeted-phage-based therapeutic strategies.33 Due to the well recognised utility of phage display technology in studying ligand – ligate interactions and in identifying a protein/peptide ligand by a process of selection rather than screening, there has been continual effort to improve the existing M13-based phage display system and to develop new systems for higher-density display of peptides and large protein domains.34 – 36 The new systems are based on large genome phage, mainly T7, T4 and lambda.9,16,17,37 A T7 display system is commercially available for display of large proteins, however, at low density. The T4 display system supports high-density display but the cloning efficiencies achieved in this system are low.37 The lambda phage has been shown to display large proteins such as b-galactosidase in functional form in about 35 copies per phage particle.16 Here also, the cloning efficiency was not very high. The system described here has simplified the process of cloning in lambda and the display density achieved for all sizes of proteins is remarkably high. In summary, the novel approach for cloning and display of peptides/large protein domains in high density on phage lambda, described here, should be useful in constructing whole genome libraries for identifying epitopes using complex polyclonal sera, isolating interacting partners and deciphering protein interaction networks with high sensitivity and specificity. Cell-targeted delivery and in vivo panning should benefit greatly from this lambda display system.

251

Functional Display of Proteins on Lambda

Materials and Methods Materials E. coli strain BM25.8 {supE thiD (lac-proAB) [F0 traD36proA þB þlacIqZDM15 ]limm434 (kanr) P1 (Cmr) hsdR (r 2m þ) (Novagen, Madison, WI) was used as Creþ host for in vivo recombination. E. coli strain TG1 (supED 2 2 (hsdM-mcrB)5(r2 [F0 traD36, k mk McrB )thiD(lac-proAB) LacIqD(lacZ)M15 ]) was used as Cre2 host for titreing phage lysates and amplification of phages. lDam imm21 nin517 was used for constructing lDL1. Collagenase was obtained from Roche Diagnostics, Germany. Anti-c-myc mAb, 9E10 was produced using hybridoma obtained from ATCC, Manassas, VA. Anti-p24 mAb, H23 was produced in-house and its epitope mapped (amino acid residues 56 – 66 of HIV-1 p24) using a phage displaybased gene-fragment library.7 GST-c-myc was produced in E. coli and purified to homogeneity by affinity chromatography.7 mAbs to PE were raised by immunising mice with a derivative of PE-38 carrying mutation in the active site. The human sera were obtained from patients undergoing immunotoxin therapy and collected after informed consent. For the present study, these samples were used as anonymous samples. HRP-conjugated antibodies were obtained from Jackson ImmunoResearch Laboratories, West Grove, PA. Construction of donor plasmid vectors The donor plasmid vector, pVCDcDL1, was assembled by ligating the following three segments of DNA bearing compatible ends. One segment was prepared by PCRbased amplification of the lambda D gene to create a HindIII site before the Shine – Dalgarno sequence and to incorporate after the last codon of D gene, a sequence encoding spacer (GGSG), followed by a collagenase cleavage site (PVGP), NheI site, ten codons of a stuffer sequence, codons for decapeptide tag, c-myc, stop codon, and SalI and EcoRI restriction sites. The assembled PCR product was digested with HindIII and EcoRI to obtain a 475 bp fragment. The second segment was also assembled by PCR and contained the origin of replication of filamentous phage ( fori) flanked by the sequence for restriction site SstI and lox P51119 on one end and the sequence for lox Pwt and an EcoRI restriction site on the other end. The product was digested with SstI and EcoRI to obtain a 515 bp fragment. The third segment formed the backbone of the plasmid vector. For this, an SstI restriction site was created by site-directed mutagenesis in pUC119 upstream of the b-lactamase gene to produce a plasmid pUCSSt. pUCSSt was digested with HindIII and Sst I and dephosphorylated to obtain a 2.5 kb fragment. pVCDcDL1 (GenBank accession no. AY10049), was obtained from ligation of the three fragments, and sequenced between HindIII and SstI sites using the dideoxy chain termination method. pVCDcDL3 (GenBank accession no. AY190304) was constructed by cloning a cassette encoding the lac promoter, RBS and the first 145 codons of lacZ flanked by SmaI/ SrfI sites, as NheI–EcoRI insert in pVCDcDL1 (Figure 1C).

lDam at map co-ordinate 24508. The ligation mix was then packaged in vitro using the Gigapack II system (Stratagene, La Jolla, CA). The phage mixture produced after packaging was plated on lawn cells (E. coli strain TG1). The plaques obtained were analysed for recombinants by PCR using primers L1 and L4, which flank the XbaI site in lambda (Figure 1B). The recombinant obtained was named lDL1. Generation of lambda cointegrates and phage production BM 25.8 cells (Creþ) and TG1 cells (Cre2) transformed with donor plasmid (carrying foreign DNA) were grown to A600 nm , 0:3 in LBAmp (LB medium containing ampicillin at 100 mg/ml) at 37 8C. Cells (1 £ 108) were harvested and suspended in 100 ml of lDL1 phage lysate at an MOI of 1.0. After incubation at 37 8C for ten minutes, the sample was diluted in 1 ml of LBAmp containing MgCl2 (10 mM) and grown at 37 8C with shaking for three hours for lysis. For large-scale recombination, the number of cells and the volume of lDL1 were increased proportionately to maintain an MOI of 1.0. The cell-free supernatant was used to infect an exponential phase culture of TG1 and Ampr colonies obtained. These Ampr colonies are immune to superinfection and carry the phage as plasmid cointegrates. The Ampr colony containing the lambda cointegrate was grown in LBAmp at 37 8C for four hours. Lambda phage are spontaneously induced in these cultures and result in complete lysis. This cell-free supernatant was then used to infect TG1 cells to obtain plaques. Phage obtained from single plaques were amplified by the liquid lysis method at an MOI of 0.01 to obtain lysate with a titre of 5 £ 109 pfu per ml. These phage were further amplified by the liquid lysis method and purified by PEG – NaCl precipitation and differential sedimentation. Construction of lambda and M13 vectors for display of various fragments of p24 DNA sequences encoding different fragments of HIV capsid protein p24 were amplified from pVCp2421029 and cloned between NheI – MluI sites to replace the stuffer fragment in pVCDcDL1 and create donor plasmids pVCDc(p241)DL1/pVCDc(p246)DL1 and pVCDc (p24)DL1. E. coli strain BM25.8 was transformed with each plasmid and recombination carried out by infecting cultures of each transformant with lDL1 phage to obtain DCO cointegrates of lDc(p241)DL1, lDc(p246)DL1 and lDc(p24)DL1 as described above. DNA encoding different p24 fragments were also cloned as Nhe I– Mlu I inserts into phagemid gIII display vector, pVC3TA726,38 and a similar phagemid gVIII display vector, pVCp240518426 (V.K.C., unpublished results) to obtain various phagemid constructs to produce phage displaying protein fused to gIIIp and gVIIIp of M13, respectively. The M13 phage displaying proteins were produced by using VCS M13 as described.39 The lambda and M13 phage were purified from cellfree supernatant by PEG precipitation followed by ultracentrifugation.

Construction of recipient lambda vector A DNA segment comprising the lac promoter, RBS and first 58 codons of lacZ flanked by sequence for lox P511 and lox Pwt was assembled by PCR to have XbaI compatible ends and ligated in the unique XbaI site in

Construction of Pseudomonas exotoxin (PE) genefragment library in l and M13 vectors Random fragments (50– 200 bp) of DNA encoding PE-38, a 38 kDa fragment of PE,23 were produced by

252

DNase I digestion and ligated as blunt-ended fragments (1 mg) in SmaI (CCCGGG)-digested pVCDcDL3 (500 ng) in the presence of restriction enzyme Srf I (GCCCGGGC) using previously described protocols.40 The ligation mix was electroporated into BM25.8 cells and plated on 150 mm LBAmpGlu (LBAmp medium containing 1% glucose) plates to obtain 5 £ 106 independent clones. The transformants were scraped and cell suspension stored at 2 70 8C. An aliquot of stored cell suspension (1 £ 108 cells) of the library was grown in 10 ml of LBAmpGlu to an A600 of 0.3. The cells were harvested and suspended in 1 ml of lDL1 phage lysate at an MOI of 1.0. After incubation at 37 8C for ten minutes, the samples were diluted in 10 ml of LBAmp containing MgCl2 (10 mM) and grown at 37 8C with shaking for three to four hours until cell lysis. The cell-free supernatant (10 ml) was used to infect an exponential phase culture of TG1 cells (10 ml) at 37 8C for ten minutes and the cell suspension was plated on 20 LBAmpGlu 150 mm plates. The Ampr colonies harbouring cointegrates were scraped and stored at 2 70 8C. Cells (1 £ 109) harbouring cointegrates were diluted into 50 ml of LBAmp medium and grown at 37 8C for eight hours to produce phage particles. The cell-free supernatant containing phage particles was directly used for affinity selection. PE-derived 50 – 200 bp DNA fragments were also ligated to SmaI-digested phagemid-based gIIIp display vector, pVCEPI13426 (V.K.C., unpublished results) to obtain the gene-fragment library in M13. A library of 6 £ 106 independent clones was obtained in TG1 cells and used to produce M13 phage displaying peptides as described.39 Construction of scFv displaying lambda phage DNA encoding the scFv fragment of the anti-mesothelin antibody, SS1 was PCR amplified using pPSC7-1-126 as template and cloned as an NheI –MluI insert in pVCDcDL1, to obtain donor plasmid pVCDcSS1DL1. BM25.8 cells were transformed with pVCDcSS1DL1 and recombination performed using lDL1 as described above to isolate a clone harbouring DCO cointegrate, lDcSS1DL1. A single colony harbouring DCO cointegrate was grown in LBAmp at 37 8C for four to six hours for lysis to occur. The supernatant was used to grow more phage by the liquid lysis method in LB medium by infecting TG1 cells at MOI 0.01. Phage from cell-free supernatant were purified by PEG – NaCl precipitation and differential sedimentation. Estimation of phage binding and affinity selection of binders by bio-panning To check the presence of binder phage, wells of microtitre plates (Maxisorp, Nunc, Rochester, NY) were coated with 1:1000 dilution of ascitic fluid of anti-c-myc mAb 9E10 and phage lysate was added to the coated wells40 and incubated for one hour at 37 8C. The unbound phage were removed by washing. To assay the captured lambda phage, 0.3 ml of exponential phase TG1 cells were added to each well and incubated for ten minutes at 37 8C. Cells were then removed and serial dilutions plated to determine phage-infected cells as pfu and cfu. The pfu and cfu indicate the number of phage bound to the coated wells. For panning of the PE gene-fragment library on mAb, wells were first coated with goat anti-mouse IgG (Fc frag-

Functional Display of Proteins on Lambda

ment-specific) antibody followed by 1:100 dilution of anti-PE mAb culture supernatant (Test wells) or buffer (Control wells). For panning of the PE gene-fragment library on human serum, wells were coated with goat anti-human (IgG þ IgM, Fc fragment-specific) antibody followed by 1:100 dilution of serum from patients treated with PE-based immunotoxins (Test wells) or pre-treatment serum of patients (Control wells). Phage lysate (1 £ 108 phages per well for lambda library and 1 £ 1010 phage per well for M13 library) was added to each well, incubated at 37 8C for one hour and unbound phages removed by washing. For the M13 library, the captured phage were eluted using low-pH buffer40 and titrated on TG1 as cfu. In the case of lambda phage, one unit of collagenase in 0.1 ml of phosphate buffer (20 mM, pH 7.4) was added to each well for ten minutes at room temperature. The released phages were titrated on TGI to obtain Ampr colonies. Individual Ampr colonies were grown and phage particles produced as described previously by infecting with helper phage for M13 clones39 and by growing colonies in LBAmp medium till complete cell lysis for lambda phage clones. The cell-free supernatants were subsequently used for ELISA. Western blot analysis and ELISA of phage For Western blots, purified phage were electrophoresed under reducing conditions on 0.1% (w/v) SDS/ 10% or 12.5% (w/v) PAG followed by electroblotting onto PVDF membrane (Immobilon, Millipore, Bedford, MA). Fusion proteins were detected with 1:1000 dilution of ascitic fluid of anti-c-myc mAb, 9E10/anti-p24 mAb, H23 followed by horse radish peroxidase (HRP)-conjugated goat anti-mouse IgG (H þ L) antibody. For ELISA, wells of Maxisorp plates (Nunc, Rochester, NY) were coated with 1:1000 dilution of ascitic fluid of mAb 9E10/H23 and purified phage were added to the coated wells. The bound phages were detected with rabbit anti-lambda polyclonal serum or rabbit anti-M13 polyclonal serum followed by HRP-conjugated goat anti-rabbit IgG (H þ L) antibody. Binding of phages produced by individual clones selected in bio-panning was tested in ELISA. For this, wells were coated with 1:1000 dilution of rabbit antilambda polyclonal serum or rabbit anti-M13 polyclonal serum and corresponding phages were added to the coated wells. After removing unbound phage, 1:100 dilution of anti-PE mAb (culture supernatant) or serum from patients treated with PE-based immunotoxins was added. The bound phage were detected with HRPconjugated goat anti-mouse IgG (H þ L) antibody or HRP-conjugated goat anti-human (IgG þ IgM) antibody. For ELISA of phages displaying SS1 scFv, microtitre wells were coated with 100 ng of recombinant mesothelin.26 After blocking the unoccupied sites with 2% non-fat dry milk, purified lambda phage were added to the coated wells and incubated at 37 8C for one hour. The unbound phage were removed by washing and the bound phage detected with rabbit antilambda polyclonal serum followed by HRP-conjugated goat anti-rabbit IgG (H þ L) antibody.

Acknowledgements The authors are grateful to Dr Ron Hoess for

253

Functional Display of Proteins on Lambda

providing l Dam vector for this work. We thank Abhishek Kulshreshta for help in panning of the M13 phage-based PE library. Dr J. P. Khurana is acknowledged for critical review of the manuscript. The Department of Biotechnology, Government of India, financially supported the work.

15.

16.

References 1. Tong, A. H., Drees, B., Nardelli, G., Bader, G. D., Brannetti, B., Castagnoli, L. et al. (2002). A combined experimental and computational strategy to define protein interaction networks for peptide recognition modules. Science, 295, 321– 324. 2. Zozulya, S., Lioubin, M., Hill, R. J., Abram, C. & Gishizky, M. L. (1999). Mapping signal transduction pathways by phage display. Nature Biotechnol. 17, 1193– 1198. 3. Zucconi, A., Dente, L., Santonico, E., Castagnoli, L. & Cesareni, G. (2001). Selection of ligands by panning of domain libraries displayed on phage lambda reveals new potential partners of synaptojanin 1. J. Mol. Biol. 307, 1329 –1339. 4. Scott, J. K. & Smith, G. P. (1990). Searching for peptide ligands with an epitope library. Science, 249, 386–390. 5. Cortese, R., Monaci, P., Luzzago, A., Santini, C., Bartoli, F., Cortese, I. et al. (1996). Selection of biologically active peptides by phage display of random peptide libraries. Curr. Opin. Biotechnol. 7, 616–621. 6. Rader, C. & Barbas, C. F., 3rd (1997). Phage display of combinatorial antibody libraries. Curr. Opin. Biotechnol. 8, 503– 508. 7. Gupta, S., Arora, K., Sampath, A., Singh, S. S., Gupta, A. & Chaudhary, V. K. (2001). Mapping of hiv-1 Gag epitopes recognized by polyclonal antibodies using gene-fragment phage display system. Prep. Biochem. Biotechnol. 31, 185– 200. 8. Kuwabara, I., Maruyama, H., Kamisue, S., Shima, M., Yoshioka, A. & Maruyama, I. N. (1999). Mapping of the minimal domain encoding a conformational epitope by lambda phage surface display: factor VIII inhibitor antibodies from haemophilia A patients. J. Immunol. Methods, 224, 89 – 99. 9. Santi, E., Capone, S., Mennuni, C., Lahm, A., Tramontano, A., Luzzago, A. & Nicosia, A. (2000). Bacteriophage lambda display of complex cDNA libraries: a new approach to functional genomics. J. Mol. Biol. 296, 497 –508. 10. Santini, C., Brennan, D., Mennuni, C., Hoess, R. H., Nicosia, A., Cortese, R. & Luzzago, A. (1998). Efficient display of an HCV cDNA expression library as C-terminal fusion to the capsid protein D of bacteriophage lambda. J. Mol. Biol. 282, 125– 135. 11. McCafferty, J., Jackson, R. H. & Chiswell, D. J. (1991). Phage-enzymes: expression and affinity chromatography of functional alkaline phosphatase on the surface of bacteriophage. Protein Eng. 4, 955–961. 12. Fuh, G. & Sidhu, S. S. (2000). Efficient phage display of polypeptides fused to the carboxy-terminus of the M13 gene-3 minor coat protein. FEBS Letters, 480, 231– 234. 13. Weiss, G. A. & Sidhu, S. S. (2000). Design and evolution of artificial M13 coat proteins. J. Mol. Biol. 300, 213–219. 14. Crameri, R., Jaussi, R., Menz, G. & Blaser, K. (1994). Display of expression products of cDNA libraries on

17. 18.

19. 20.

21.

22.

23.

24.

25.

26. 27.

28.

29.

30.

phage surfaces. A versatile screening system for selective isolation of genes by specific gene-product/ligand interaction. Eur. J. Biochem. 226, 53 – 58. Dunn, I. S. (1995). Assembly of functional bacteriophage lambda virions incorporating C-terminal peptide or protein fusions with the major tail protein. J. Mol. Biol. 248, 497– 506. Mikawa, Y. G., Maruyama, I. N. & Brenner, S. (1996). Surface display of proteins on bacteriophage lambda heads. J. Mol. Biol. 262, 21 – 30. Sternberg, N. & Hoess, R. H. (1995). Display of peptides and proteins on the surface of bacteriophage lambda. Proc. Natl Acad. Sci. USA, 92, 1609– 1613. Sheets, M. D., Amersdorfer, P., Finnern, R., Sargent, P., Lindquist, E., Schier, R. et al. (1998). Efficient construction of a large nonimmune phage antibody library: the production of high-affinity human single-chain antibodies to protein antigens. Proc. Natl Acad. Sci. USA, 95, 6157– 6162. Hoess, R. H., Wierzbicki, A. & Abremski, K. (1986). The role of the loxP spacer region in P1 site-specific recombination. Nucl. Acids Res. 14, 2287– 2300. Georgopoulos, C., Tilly, K. & Casjens, S. (1983). Lambda II (Hendrix, R. W., Roberts, J. W., Stahl, F. W. & Weisberg, R. A., eds), Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY. Gamble, T. R., Yoo, S., Vajdos, F. F., von Schwedler, U. K., Worthylake, D. K., Wang, H. et al. (1997). Structure of the carboxyl-terminal dimerization domain of the HIV-1 capsid protein. Science, 278, 849– 853. Gitti, R. K., Lee, B. M., Walker, J., Summers, M. F., Yoo, S. & Sundquist, W. I. (1996). Structure of the amino-terminal core domain of the HIV-1 capsid protein. Science, 273, 231– 235. Debinski, W. & Pastan, I. (1992). Monovalent immunotoxin containing truncated form of Pseudomonas exotoxin as potent antitumor agent. Cancer Res. 52, 5379 –5385. Pai, L. H., Bookman, M. A., Ozols, R. F., Young, R. C., Smith, J. W., 2nd, Longo, D. L. et al. (1991). Clinical evaluation of intraperitoneal Pseudomonas exotoxin immunoconjugate OVB3-PE in patients with ovarian cancer. J. Clin. Oncol. 9, 2095– 2103. Chowdhury, P. S., Viner, J. L., Beers, R. & Pastan, I. (1998). Isolation of a high-affinity stable single-chain Fv specific for mesothelin from DNA-immunized mice by phage display and construction of a recombinant immunotoxin with anti-tumor activity. Proc. Natl Acad. Sci. USA, 95, 669– 674. Chowdhury, P. S. & Pastan, I. (1999). Improving antibody affinity by mimicking somatic hypermutation in vitro. Nature Biotechnol. 17, 568– 572. Niwa, M., Maruyama, H., Fujimoto, T., Dohi, K. & Maruyama, I. N. (2000). Affinity selection of cDNA libraries by lambda phage surface display. Gene, 256, 229 –236. Rose, S., Hensley, P., O’Shannessy, D. J., Culp, J., Debouck, C. & Chaiken, I. (1992). Characterization of HIV-1 p24 self-association using analytical affinity chromatography. Proteins: Struct. Funct. Genet. 13, 112 – 119. Gupta, S., Arora, K., Gupta, A. & Chaudhary, V. K. (2000). Gag-derived proteins of HIV-1 isolates from Indian patients: cloning, expression, and purification of p24 of B- and C-subtypes. Protein Expr. Purif. 19, 321 –328. Forrer, P. & Jaussi, R. (1998). High-level expression of soluble heterologous proteins in the cytoplasm of

254

31.

32.

33.

34.

35. 36.

Functional Display of Proteins on Lambda

Escherichia coli by fusion to the bacteriophage Lambda head protein D. Gene, 224, 45 – 52. Kuwabara, I., Maruyama, H., Mikawa, Y. G., Zuberi, R. I., Liu, F. T. & Maruyama, I. N. (1997). Efficient epitope mapping by bacteriophage lambda surface display. Nature Biotechnol. 15, 74 – 78. Bessette, P. H., Aslund, F., Beckwith, J. & Georgiou, G. (1999). Efficient folding of proteins with multiple disulfide bonds in the Escherichia coli cytoplasm. Proc. Natl Acad. Sci. USA, 96, 13703– 13708. Larocca, D., Burg, M. A., Jensen-Pergakes, K., Ravey, E. P., Gonzalez, A. M. & Baird, A. (2002). Evolving phage vectors for cell targeted gene delivery. Curr. Pharm. Biotechnol. 3, 45 – 57. Rondot, S., Koch, J., Breitling, F. & Dubel, S. (2001). A helper phage to improve single-chain antibody presentation in phage display. Nature Biotechnol. 19, 75 – 78. Sidhu, S. S., Weiss, G. A. & Wells, J. A. (2000). High copy display of large proteins on phage for functional selections. J. Mol. Biol. 296, 487– 495. Soltes, G., Barker, H., Marmai, K., Pun, E., Yuen, A. & Wiersma, E. J. (2003). A new helper phage and

37.

38.

39.

40.

phagemid vector system improves viral display of antibody Fab fragments and avoids propagation of insert-less virions. J. Immunol. Methods, 274, 233– 244. Ren, Z. J., Lewis, G. K., Wingfield, P. T., Locke, E. G., Steven, A. C. & Black, L. W. (1996). Phage display of intact domains at high copy number: a system based on SOC, the small outer capsid protein of bacteriophage T4. Protein Sci. 5, 1833– 1843. Sampath, A., Abrol, S. & Chaudhary, V. K. (1997). Versatile vectors for direct cloning and ligationindependent cloning of PCR-amplified fragments for surface display on filamentous bacteriophages. Gene, 190, 5 –10. Kushwaha, A., Chowdhury, P. S., Arora, K., Abrol, S. & Chaudhary, V. K. (1994). Construction and characterization of M13 bacteriophages displaying functional IgG-binding domains of staphylococcal protein A. Gene, 151, 45 – 51. Gupta, S., Arora, K., Sampath, A., Khurana, S., Singh, S. S., Gupta, A. & Chaudhary, V. K. (1999). Simplified gene-fragment phage display system for epitope mapping. Biotechniques, 27, 328– 330.

Edited by M. Gottesman (Received 4 June 2003; received in revised form 29 August 2003; accepted 15 September 2003)