Immunogenicity and epitope mapping of foreign sequences via genetically engineered filamentous phage.

Repeat regions of the circumsporozoite protein gene of Plasmodium falciparum were cloned into the pIII gene of a filamentous phage. These genetically engineered filamentous phage display the recombinant proteins on their surface. We demonstrate that they are both antigenic and immunogenic in rabbits. The recombinant phage were shown to be useful as a source of antigen for this scarce malaria protein, for producing carrier-hapten conjugates for obtaining immunological reagents in rabbits, and for B epitope mapping. In addition, in mice the antibody response to the cloned antigens seems to be controlled by immune response genes. Therefore this system also has the potential for use in helper T cell epitope mapping using inbred mouse strains. This advantage will be of use in vaccine development.

Repeat regions of the circumsporozoite protein gene of Plasmodium falciparum were cloned into the pIII gene of a filamentous phage. These genetically engineered filamentous phage display the recombinant proteins on their surface. We demonstrate that they are both antigenic and immunogenic in rabbits. The recombinant phage were shown to be useful as a source of antigen for this scarce malaria protein, for producing carrier-hapten conjugates for obtaining immunological reagents in rabbits, and for B epitope mapping. In addition, in mice the antibody response to the cloned antigens seems to be controlled by immune response genes. Therefore this system also has the potential for use in helper T cell epitope mapping using inbred mouse strains. This advantage will be of use in vaccine development.
The generation of antisera to defined peptide sequences usually involves the chemical synthesis of a peptide and its conjugation to a carrier molecule. Here we present the use of genetically engineered filamentous phage as a means of obtaining both antigen and carrier for antibody production and source of antigen for immunological studies, including epitope mapping.
The system depends on the construction of a fusion protein by inserting DNA fragments into the pIII minor coat protein of the bacteriophage fl. The molecular biology and virology of this system have been studied extensively (1)(2)(3)(4)(5). The minor coat protein is located at one end of the filamentous bacteriophage (M13, fl, fd) (1). DNA fragments inserted into the pIII gene are expressed, provided the reading frame of the gene is maintained (5). Insertions that do not maintain the reading frame thus result in uninfective phage. The minor coat protein is accessible to antibodies (i.e. is antigenic), but little is known about the immunological response (ie. immunogenicity) of mammals to recombinant phage, since a measure of antibody binding to an antigen does not have predictive value for how useful the antigen will be in stimulating an antibody response.
Several factors led us to investigate the immunology of this phage system. It is a system that is analogous to making other recombinant proteins but has the distinct advantage that the * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.  recombinant protein is automatically purified. This occurs because the fusion protein is part of a nonlysogenic bacteriophage virion (the male-specific Escherichia coli fl phage) which is extruded by the cell into the culture supernatant, thus making the purification procedure simple. Furthermore, in contrast to chemical synthesis this system leads to essentially pure products since the fidelity of translation approaches 100%. Therefore, as the system appeared to be an ideal immunological tool, we investigated the immune response of mice and rabbits to the recombinant phage. We measured antibody production in response to the injection of different recombinant phage and the genetic restriction of that response. Recombinant phage were found to elicit responses to the fusion protein in one genetic background of mice, but not in another (i.e. it is genetically restricted), and in rabbits. The phage were also shown to be useful for the localization of B epitopes on proteins.

MATERIALS AND METHODS
Source of DNAs and Enzymes-Fragments of the Plosmdium fakiparum 7G8 circumsporozoite (CS)' protein gene were either synthesized using phosphoramidite chemistry on a DNA synthesizer (Applied Biosystems, Inc.) or derived from the clone, XmPfl (6). All restriction endonucleases, Klenow fragment, and T4 polynucleotide kinase were purchased from Bethesda Research Laboratories. Calf intestine alkaline phosphatase was purchased from Boehringer Mannheim.
Cloning of CS Protein Fragments into fl Replicative Form DNA-The major tetramer encoding repeat (NANP) of the CS protein was cloned into the unique BamHI site of f l bacteriophage using two synthesized complementary oligonucleotides (sense: 5'-GAACGCGAACCCG-3'; anti-sense: 5"GATCCGGGTTCGCGTTencoded amino acid sequence: DP(NANP),DP). The synthesized oligonucleotides were purified from contaminants by denaturingpolyacrylamide gel electrophoresis. The least mobile bands were excised, and the material was eluted into 0.5 M ammonium acetate by overnight shaking at 37 "C. The eluted oligonucleotides were passed through a Sep-Pak CIS cartridge (Millipore, Waters Associates), kinased, and annealed (100 mM NaCl, 50 mM Tris-HC1 (pH 7.5)) by mixing equal amounts (5 pg) and heating to 65 "C for 10 min, followed by gradual cooling to room temperature. The annealed oligonucleotides were then ligated into BamHI-linearized and dephosphorylated fl replicative form DNA. An alternating series of major (NANP) and minor (NVDP) tetramer repeats were also cloned into f l bacteriophage by first isolating the 7G8 CS protein gene insert from a subclone of XmPfl and digesting it to completion with Sau3A restriction endonuclease. Resultant fragments were randomly cloned into the BarnHI-linearized and dephosphorylated fl replicative form DNA.

CGGATTCGCATTTGGGTTTGCGTTTGGATTTGCATTAG-3';
Epitope Mapping Using Filamentous Phage 4319 cells using E. coli K91 cells for lawn production. DNA Sequencing-Phage DNAs were sequenced by the dideoxy procedure (7) for confirmation of DNA insertion. An oligonucleotide sequencing primer (5'-GGTCAGACGATTGGCC-3') was synthesized for this purpose. This primer is complementary to the sequence at position 2245-2260 and is 20 nucleotides downstream (ie. 3') of the unique BamHI site in the pIII gene of f l (8).
Rabbit Immunizations-Rabbit immunizations were carried out following standard protocols. Phage used as antigens for immunizations were purified by a modification of the "purification I" procedure of Lin et al. (9). In brief, culture supernatants (200-500 ml) were polyethylene glycol-precipitated (final concentration 4% polyethylene glycol-8000, 0.5 M NaCI), and the phage pellet was resuspended in 0.01 M Tris-HC1, pH 7.6, 1.0 mM EDTA, 0.1% Sarkosyl and incubated for 30 min at room temperature. The material was precipitated by polyethylene glycol, the pellet resuspended in several milliliters of 0.1 M NaCI, 1.0 mM EDTA, 0.01 M Tris-HC1 (pH 7.6) (NET), and subjected to cesium chloride centrifugation for 18 h at 110,000 X g (0.386 g of CsCl/ml of buffer, 40 ml total). The visible phage bands were tapped by syringe and diluted with NET. The phage were pelleted by centrifugation at 200,000 X g for 3 h and finally resuspended in PBS. Rabbits were immunized by subcutaneous subscapular injections on day 0 (0.1 mg in CFA) and boosted on days 14 and 28 (0.1 mg in IFA). The animals were bled on days 21 and 35. All bleeds were clotted at room temperature, centrifuged, and filtered (0.45 pm).
Mouse Immunizations-BALB/c (H-2d) and C57BL/10 (H-2b) mice were immunized intraperitoneally following standard protocols. The initial immunization was done with 0.1 mg of purified phage (assuming 1 mg/ml = 1 OD, ) in PBS with CFA (1:l) in 100 p1 total volume. Subsequent boosts, consisting of 50 pg phage with IFA in a total volume of 50 pl, were administered every 2 weeks thereafter. Serum samples were obtained by bleeding from the tail artery before each boost.
Dot Assays-DNA dot blots were prepared by spotting 5 pl of culture supernatant on nitrocellulose paper prewetted in 10 X SSC (1 X SSC = 0.3 M NaCI, 0.03 M sodium citrate), processed as described (lo), and probed with nick-translated (11) CS protein gene insert. Immunodot blots were carried out by spotting 5 pI of culture supernatant and were processed as described using a 1:10,000 dilution of a mixture of three monoclonal antibodies (236.4, 4D9.1, 5G5.3) which are specific for the P. fakiparum CS protein repeats (6). A second antibody (rabbit anti-mouse IgG) was used for amplification of signals at 1:500 dilution. Antibodies bound to the filter were detected by incubation with 1 pCi of '*T-labeled protein A followed by autoradi-Enzyme-linked Immumsorbent Assays-Enzyme-linked immunosorbent assays were conducted as follows: antigens (per well, 0.1 mg of fl phage, 0.1 pg of R32tet32 (i.e. a fusion protein consisting of 32 tetramers from P. fakiparum followed by 32 amino acids of a tetracycline resistance gene read out of frame; Smith, Kline and French; BPP23, JD9908289) (12) or 0.1 pg (NANP), conjugated to Keyhole Limpet Hemocyanin were adsorbed onto Immulon-1 microtiter plates (Dynatech Laboratories) in a volume of 100 pl of PBS overnight at 4 "C. The wells were emptied and blocked with 5% non-fat dry milk (Carnation Co., Los Angeles, CA), 0.05% Tween 20 (Sigma), and PBS (250 /*]/well) for 2 h. After washing the wells 3 or 4 times with PBS/ Tween, serial dilutions of sera (100 pl, in blocking buffer) were added to the wells and incubated for 2 h. The wells were washed 3 or 4 times, after which a second antibody was added (horseradish peroxidase-conjugated goat anti-rabbit Ig, Cooper Biomedical) in blocking buffer (1:lOOO dilution) and incubated for 2 h. The wells were again washed before the color-developing reagents were added (Peroxidase Substrate System ABTS, Kirkegaard and Perry Laboratories, Inc.). Color development was monitored, after approximately 1-h incubation, on a Titertek Multiskan MC reader and is presented as absorbance at 414 nm wavelength.

RESULTS
Cloning of CS Protein Tetramer Repeats-Recombinant filamentous phage containing P. falciparum CS protein re-peats (6) were constructed using synthetic oligonucleotides (clone fPfl6) and restriction enzyme fragments (clones fPfl through P f l 2 ) . All plaques picked were checked for insertion of the synthetic tetramer or gene fragment insert by hybridization with the 7G8 CS protein gene or synthetic oligonucleotide (data not shown). Sequence analysis of these clones confirmed the hybridization results, and clones identified as having inserts in the proper orientation were used for further experiments. Of the gene fragments generated by restriction enzyme digestion, only several were predicted to be capable of maintaining the open reading frame of the pIII gene. These fragments encoded the tetramer major repeat, NANP, and the variant repeat, NVDP. The insert sequences of these clones are shown in Table I.
Immunoekctron Microscopy-The antigenicity of the fusions was tested as follows. A recombinant phage (fPfl6) containing the synthetic gene fragment was purified and analyzed by EM (Fig. 1). Mouse antisera against a synthetic (NANP)6 peptide did not bind to fl virions (Fig. 1A). The same sera were specific against one end of the fPfl6 virions (Fig. 1B). This is consistent with the location of pIII (3, 4, 13), into which the P. fakiparum tetramer repeats were cloned. Normal mouse (i.e. preimmune) sera showed no reaction against the fPfl6 virion (Fig. IC).
Determination of Antigenic Sites (B Epitope Mapping)-A protein immunodot blot assay (Fig. 2), using a mixture of three mAbs against P. fakiparum CS protein, showed that only those clones with four or more tetramer repeats (i.e two or more of the small Sau3A fragments, Table I) reacted with the mAbs and that the incorrect insert orientation was not reactive.
Generation of Serological Reagents-The immunogenicity of the fusion proteins was tested as follows. Rabbits were immunized with the fl phage or against one of the recombinant phage (fPf2) (Table I), with and without adjuvant (CFA, IFA). The serum from all rabbits had high antibody titers to the fl virion proteins, and only the rabbits immunized against the recombinant phage had antibodies against the R32tet32 protein ( Fig. 3 and data not shown). The use of an adjuvant increased the apparent response to the repeats. All animals developed antibodies against fl proteins, whereas rabbits immunized with fPf2 and fPfl6 had anti-repeat titers (data not shown). In a separate experiment, rabbits were immunized with fl, fPf2, or fPfl6 using CFA. No anti-repeat antibodies   Table I. were found in the anti-fl sera, but significant anti-repeat titers were found in the anti-fPf2 and anti-fPfl6 sera (data not shown). Significant titers against the fl proteins were present in all three sera.
Inbred mice immunized with these phage displayed different responses. For example, no significant anti-repeat titers were obtained in BALB/c mice (primary bleed, Fig. 4), even after three boosts (secondary bleed, Fig. 4, and data not shown) and despite significant boosting of anti-fl titers. In contrast, anti-repeat titers were observed in C57BL/10 mice after the initial immunization (primary bleed, Fig. 4). These titers increased upon boosting (secondary bleed, Fig. 4). Antifl phage antibodies were present in all immunized mice. We have found that comparable antibody titers can be achieved using recombinant phage as with conjugated peptides, but in one case a conjugated peptide was not immunogenic, whereas the same sequence expressed on a recombinant phage was immunogenic.2

DISCUSSION
We describe the immunological response to recombinant proteins produced in a bacterial expression system in which the fusion protein is expressed on a viral surface and is automatically biologically enriched. Epitope expression is accomplished by cloning into the unique BamHI restriction enzyme site of a minor coat protein (pIII) gene of the filamentous bacteriophage, fl (5). The cloning site is located between the two functional domains of the protein. The Nterminal portion of pIII binds to the E. coli F pilus during infection, while the C-terminal portion is buried in the virion and functions in morphogenesis (15). Insertion of foreign sequences between these two domains is possible without significantly affecting either infectivity or morphogenesis. The maximum length of the fragments that can be inserted is unknown, but the largest insert reported to date encodes 57 amino acids (5). With respect to epitope mapping as described here, this is more than sufficient to code for any linear epitope.
Using this system we cloned the repeats of the circumsporozoite protein of the human malaria parasite P. falciparum.
In agreement with previous work, the repeats on the recombinant phage were accessible to antibody (i.e. antigenic) as evidenced here by EM. We then used this system for B epitope  mapping and confirmed that the mAbs used here recognize three or more repeats (16,17). This was evident from the result that only those recombinant phage with at least three repeats reacted with the mAbs. This system is thus amenable to rapid B epitope mapping, especially if one uses gene fragments derived by various DNA fragmentation procedures (e.g. random DNase (18) sonication (19)) where the antibody of interest can be used to select the epitope-bearing phage. Rapid DNA sequencing can then be used to identify the inserted DNA fragments. B epitope mapping with this system using other proteins of parasitological interest is currently in progress.
Despite the low mass representation of the repeats as a result of cloning into a minor coat protein (there are about 5 copies of pIII per virion; repeats in fPf2 represent 0.1% of total virion protein), there was a sufficient response to the repeats when rabbits were immunized with recombinant phage preparations. Thus, the repeats cloned in these recombinant phage are not only antigenic, but immunogenic in rabbits. However, responses in inbred mice varied, indicating that the system may also be amenable to helper T cell epitope mapping. We are currently using this system for mapping of responses to mouse malaria antigens as well as a source of immunogen for vaccine development studies.
From our immunization experiments carried out in inbred mice it was clear that despite the antigenicity of the phage, the phage were not necessarily immunogenic. Since no antirepeat antibodies were generated in BALB/c mice, it appears that there are no helper T cell epitopes on pIII that are recognized in the BALB/c H-2d genetic background for T cell-dependent (ie. helper) anti-repeat antibody production. This conclusion is based on the antibody responses of the same phage constructs in C57BL/10 mice and the observation that the P. fakiparum NANP tetramer repeats are recognized as helper T cell epitopes only in the context of the I-Ab gene (e.g. the C57BL/10 H-2" genetic background) (14,20). In contrast to the results in the BALB/c mice, the C57BL/10 mice did respond to the repeats on the phage. This is consist- Log2 Reciprocal Dilution ent with mapping experiments that indicate that mice of the H-2d genetic background (e.g. BALB/c, B10.D2), if immunized with the NANP repeats alone, will not generate antibodies to the repeats (14). A single NVDP variant repeat within a series of NANP repeats are also not similarly recognized in the H-2d background. However, alternating NANP and NVDP repeats, which are present on the native circumsporozoite protein, have yet to be tested. In this context, it should be noted that the C57BL/10 mice responded to not just the NANP repeats ( f P f l G ) , but also to the alternating repeats (fPf2). It is therefore not known if antibody titers to the fPf2 recombinant are due to the H-2" recognition of the alternating repeats or if it is due to an H-2" restricted recognition of an epitope(s) on the pIII protein itself. Such a response is possible since linking of a helper T site to a B epitope can lead to an immune response. This was recently demonstrated by linking the repeats to an identified CS protein helper T cell epitope (labeled ThBR, and recognized by the I-Ak allele), which resulted in the production of anti-repeat antibodies by otherwise low responding mice (21). Efforts toward inserting known genetically restricted helper T sites into the fl genome for the purpose of ensuring an immunological response are currently in progress.
Thus, the recombinant filamentous phage approach to obtaining specific peptide antigens has the following advantages over chemical synthesis: 1) the products obtained are the result of the biological fidelity of translational machinery and is not subject to the 70-94% purity levels common in the solid phase synthesis of peptides (22), which may be critical for some work (e.g. epitope mapping); 2) conjugation of the peptide to carrier molecules is not necessary, since the product is expressed as a fusion protein; 3) extensive purification procedures are not necessary because the biology of the system allows for self-enrichment (although we used extended procedures, polyethylene glycol precipitation may suffice for most work); 4) the phage represent an easily renewable source of antigen, since more material can be easily obtained by growth of bacterial cultures; 5 ) cost is minimal. Also, should peptides Epitope Mapping Using Filamentous Phage free of fusion sequences be needed, the system may be ame-