An epitope fragment library from the influenza HA gene (influenza X-31 [H3N2], 15) was generated by randomly digesting the HA DNA with DNAse I. The conditions were adjusted in order to create predominantly 50 – 100 bp fragments that were subsequently cloned into the modified pORFES vector 12. The "empty" pORFES IV vector contains an ompA leader followed by the β-lactamase gene, which is out of frame as shown in Figure 1. Insertion of DNA fragments at the NaeI site that restore the reading frame between the ompA leader and the β-lactamase coding sequence, permits secretion and allows for positive selection using β-lactam antibiotics. E. coli XLOLR electrocompetent cells were transfected with the ligation products. After 1-hour recovery at 37°C, aliquots of the cells were plated on LB agar plates containing either chloramphenicol alone or in combination with carbenicillin. Since cell growth in the presence of carbenicillin requires restoration of the β-lactamase reading frame, colonies growing on chloramphenicol alone served as a representation of the total number of transformation events regardless of the presence or nature of the insert (i.e., containing either the self-ligated vector without an insert, or the vector ligated with an insert but not necessarily in-frame with β-lactamase). As expected, approximately 1/3 of the transformation events resulted in restoring the correct reading frame. The size of the primary HA library in pORFES was determined by plating on agar plates containing carbenicillin and chloramphenicol.

Approximately 1.7 × 10⁶ independent clones were generated. Since the insert could have one of two possible orientations in three reading frames, and since some of these fragments may have included stop codons, theoretically less than 1/6 of all clones should encode cognate HA peptides. Based on this assumption, the library size for cognate HA peptides was estimated to contain approximately 2.8 × 10⁵ independent clones.

Amplification of the HA fragment library in pORFES IV permitted the positive selection of open reading frames and provided ample material for directional cloning into the display vector JC-M13-88. The unique restriction sites, XbaI-HindIII, facilitated directional transfer of the library from pORFES IV into JC-M13-88. These XbaI-HindIII fragments contain the ribosome-binding site, the ompA leader sequence followed by the random insert of the library. Although the DNA digestion yielded mainly 50 – 100 bp fragments, larger fragments were also found in the pORFES vector. In order to obtain only peptides of a length of about 15 to 30 amino acids, the XbaI-HindIII fragments were resolved on a TBE gel and DNA fragments with the appropriate size eluted. Cloning the libraries into JC-M13-88 afforded 6 × 10⁶ independent clones for the HA phage library. This figure is ten-fold higher than the original pORFES IV HA library size. This over-representation ensures that most of the library members are transferred to the display system.

The well-characterized mAb 12CA5 that binds to the HA sequence YPYDVPDYAS 16 was used to screen the initial HA library prior to panning in order to determine the initial frequency of the cognate peptide fragment. Approximately 4100 plaques of the naïve (i.e. prior to panning) library were analyzed by probing filter lifts with mAb 12CA5. Initially, only 0.4% plaques were immunoreactive. This number increased with each round of selection. The library was subjected to 3 rounds of panning against mAb 12CA5. The second round of panning resulted in a 12,000-fold increase of eluted phage (2300 fold over background). Approximately 18% of the plaques stained strongly with mAb 12CA5. This number increased to 42% after the third round of selection (Figures 2, 3, 4). Phage from 6 stained plaques were isolated and the insert DNA sequence determined. The results are shown in Table 1 (see additional file 1). Half the clones were identical whereas the other 3 clones differed from each other. All 6 clones contained the YPYDVPDYAS motif.

Subsequently, the library was evaluated using a rabbit polyclonal IgG raised against the peptide CKRGPDSGFFSRLNWLYKSG. When 2400 plaques of the initial library were screened by filter lift for immunoreactive clones, a higher background staining was observed relative to filter lifts stained with mAb 12CA5. However, one plaque was found to stain strongly with the pAb 07431. During selection, the number of eluted phage increased with each round. After the third round, 91% of plaques bound to pAb 07431 as determined by filter lifts (Figures 5, 6, 7). A sequence analysis of 16 clones revealed only 3 different sequences as shown in Table 2 (see additional file 2). All of the selected sequences contained the GFFSRLNWLTKS motif that is part of the peptide used for the immunizations. The immunizing peptide was derived from influenza strain X47 (HA1, 17) and differed by a single substitution of T to Y from the gene used for this study (derived from influenza X-31 [H3N2]) as shown in Table 2. Despite this point variation the approach permitted recovery of the epitope.

Interestingly, though, clones TSS 399 and 401–403 encoded additional sequences resulting from frame shifts and a multiple insertion event. A BLAST search of the nucleotide sequence encoding the TSS 399 peptide on phage (5' – TCA TGT GGG CCT GCC AGA GAG GCA ACA TTA GGT GCA ACA TTT GCA TTT GAG TGT GGT AGC GGT TTT TTC AGT AGA CTG AAC TGG TTG ACC AAA TCA-3') in GenBank (access code V01103), revealed that nucleotide1–54 (shown in bold) aligned with nucleotides 1681–1734 of HA (incorrect frame). As expected, a search with the corresponding encoded 18-mer peptide (SCGPAREATLG ATFAFEC) failed to align on the HA polypeptide. Nucleotides 57–96 encoding the cognate epitope aligned with HA nucleotides 509–548 (in frame).

The mAb 12CA5 specific for a HA peptide (YPYDVPDYAS)16 and the rabbit polyclonal serum 07431 against the peptide (CKRGPDSGFFSRLNWLYKSG) derived from Influenza X-47 HA1 were kindly provided by Dr. R. A. Lerner (The Scripps Research Institute). The rabbit IgG fraction was purified from the sera using protein A-sepharose (Pierce, Rochford, IL, USA).

The plasmid Open Reading Frame Expression and Secretion IV is based on(pORFES) 1214 with an insert modification shown in Figure 1. The phage display vector JC-M13-88 has been described elsewhere 121314. The host strains Escherichia coli XLOLR and XL1-Blue were purchased from Stratagene (La Jolla, CA, USA). Plasmid pCMU encoding HA cDNA (derived from influenza X-31 [H3N2], 15) was a kind gift from Dr. G. Aichinger (Hammersmith Hospital, London, UK). All enzymes were purchased from Boehringer-Mannheim (Indianapolis, IN, USA).

The plasmid pCMU (100 μg) was digested with SmaI/XbaI to release the HA coding insert. The gel-purified inserts (~10 μg DNA) was digested with 70 ng of DNAse in 50 mM Tris-HCl, pH 7.2, containing 40 mM MnCl² (130 μl) for 4 minute sat 37°C and the reaction was terminated by adding EDTA to a final concentration of 70 mM. After precipitation, the fragments were treated with 0.001 units mung bean nuclease per μg DNA for 5 minutes at 37°C. The resulting mixtures of blunt ended fragments were ligated between the ompA leader sequence and the β-lactamase gene into pORFES IV that had been digested with NaeI and treated with calf intestine phosphatase. The ligation products were transfected into non-suppressing E. coli XLOLR via electroporation and propagated overnight in Super Broth containing 100 μg/ml carbenicillin at 37°C. The inserts that restore the reading frame and that are readily translocated into the periplasm along with the fused β-lactamase confer β-lactam antibiotic resistance. The transfection efficiency was monitored by plating aliquots on agar plates containing 25 μg/ml of chloramphenicol with or without 100 μg/ml carbenicillin. The pORFES IV library DNA was recovered from the overnight culture and inserts released by digestion with XbaI and HindIII. The fragments of 160–210 bp were resolved on a 3% agarose TBE gel.

The 160–210 bp XbaI/HindIII fragments encode a ribosome binding site, the ompA leader and the HA derived inserts. These were directionally cloned into the phage vector JC-M13-88, between a lac promoter and a synthetic copy of gene VIII. The ligation products were transfected into XL1-Blue E. coli cells via electroporation and the transfection efficiency monitored by plaque assay. The phage were propagated for 16 h at 37°C in Super Broth containing 1 mM isopropyl-β-D-thiogalactopyranoside (IPTG) and isolated by PEG precipitation twice following standard methods 20. The phage pellets were resuspended in phosphate buffered saline (PBS) and stored at 4°C.

The JC-M13-88 library containing the HA gene fragments was panned against mAb 12CA5 and polyclonal rabbit IgG 07431. Multi-well maixisorp strips (Nalge Nunc International, Naperville, IL USA) were coated with 100 μl of a 8 μg/ml antibody solution in PBS or with 5 mg/ml BSA in PBS (negative control) over night at 4°C. The wells were washed 3 times with PBS containing 0.1% Tween (PBST), blocked with 0.5% BSA in PBS for 1 h at 37°C and washed as before. Approximately 10¹⁰ plaque forming units (pfu) were added in 100 μl of PBS containing 0.25% BSA and 0.05% Tween. The panning was carried out in duplicate. After incubation for 90 minutes at 37°C, the wells were washed 10 times with PBST, the phage eluted with 100 μl of 100 mM glycine, pH 2.2, at ambient temperature for 10 minutes and the solution neutralized by adding 10 μl of 1 M Tris solution (pH not adjusted). An aliquot of the solution was retained to assess the phage output and the rest was used to infect E. coli XL1-Blue cells for propagation as described above.

To determine the relative abundance of individual binding clones, filter lifts were carried out on the library prior to panning, and subsequently after each round of panning. Phage plaques were overlaid with nitrocellulose filters for 10 minutes at 37°C. The filters were subsequently removed and blocked with 0.5% BSA in PBS for 1 h at RT and washed 3 times with PBST. The mAb 12CA5 and pAb 07431 were adjusted to 2 μg/ml PBST and the filters incubated with the Ab solution for 1 h at RT. The filters were washed as before and incubated with a goat anti-mouse antibody (Southern Biotechnology Associates, Inc., Birmingham, AL USA), or a goat anti-rabbit IgG F(ab)2 fragment (United States Biochemical Corp., Cleveland, OH, USA) conjugated to alkaline phosphatase. Both secondary antibodies were diluted in PBST(1 μg/ml final concentration) and the filters incubated for 1 h at RT. After washing as before, the filters were stained with 5-bromo-4-chloro-3-indolylphosphate (BCIP) and 4-nitro blue tetrazolium chloride (NBT).

Individual phage plaques were picked and grown overnight after infection with E. coli XL1-Blue cells 20. The double stranded replicative form (RF) DNA template was sequenced.