The sera samples used in this study were provided by patients at Taipei Medical University-Wan Fang Medical Center (TMU-WFMC). Sera were prepared from convalescent SARS-CoV-infected patients and apparently healthy individuals. All collected sera were screened by a one-step anti-SARS coronavirus test kit (SD Inc., Korea) to confirm the presence or absence of anti-SARS-CoV antibodies.

SARS-CoV genomic RNA was directly extracted from the sputum sample of a hospitalized patient, diagnosed as SARS-CoV infected at TMU-WFMC, as described by the manufacturer with minor modification (Trizol reagent, Invitrogen). The whole procedure was performed in a biosafety level (BSL) 2 laboratory to avoid any potential contamination. In brief, 2 ml of Trizol reagent was added to a 0.2 ml sputum sample and stored at 4 °C until used. One millilitre of virus-inactivated sample was transferred to a 1.5 ml Eppendorf tube containing 0.2 ml of chloroform. After vortexing and centrifuging at 10,000 rpm for 10 min at 4 °C, the colorless upper aqueous phase was collected and mixed with an equal volume of isopropanol to precipitate the total RNAs. We kept the mixture at room temperature (RT) for 10 min and recentrifuged it at 10,000 rpm for 10 min at 4 °C. A gel-like pellet containing RNA was formed, washed once with 1 ml of 75% ethanol and centrifuged at 10,000 rpm for 5 min. After removing the supernatant, the pellet was air-dried and dissolved in 15 μl of RNase-free water. The extracted RNA sample was used immediately for viral gene amplification.

To amplify truncated fragments of S gene, we designed 10 sets of primers according to the published nucleotide sequences of the SARS-CoV RNA genome (GenBank Accession NC_004718). Conditions for SARS-CoV RNA amplification were performed using a one-step RT-PCR, kit as described by the manufacturers (Qiagen, Valencia, CA, USA). In brief, 5 μl of the isolated total RNAs was used as template and added to 45 μl of a mixture of 26 μl RNase-free water, 10 μl 5× one-step RT-PCR buffer, 2 μl dNTPs mixture (10 mM/each), 2 μl one-step RT-PCR enzyme mixture, 2.5 μl/primer (10 μM/each). The amplification program consisted of a reverse-transcription reaction of 50 °C for 30 min and 95 °C for 15 min, 40 cycles of 94 °C for 30 s, 55 °C for 30 s and 72 °C for 90 s, followed by 72 °C for 10 min. The RT-PCR products were kept at −20 °C until used.

We amplified 10 truncated S-gene fragments of 300–750 bp in length with PCR, treated them with BamHI and XbaI and ligated them into a pET-21 expression vector (Novagen, Germany). The resultant plasmids were transformed into an E. coli BL-21 (DE3) strain. We grew clones with the respective insert in 5 ml LB medium containing ampicillin (100 μg/ml) at 37 °C overnight. The bacterial culture was 10-fold diluted in the same LB medium and grown for an additional 3–4 h until OD₆₀₀ reached 0.6. To induce his-fused S protein expression, the iso-propyl-β-d-thiogalactopyranoside (IPTG) was added to a final concentration of 0.5 mM in the culture, which was further incubated for 4 h. After centrifugation, the cell pellet was resuspended in 2 ml of binding buffer (100 mM NaCl, 10 mM Tris–Cl, and 8 M urea pH 8.0) and lysed by three cycles of freezing (−70 °C) and thawing (37 °C). After centrifugation at 12,000 rpm for 10 min at 4 °C, the resulting cellular lysate was incubated with Ni²⁺-charged sepharose for protein purification according to the manufacturer's instructions (Amersham Biosciences, Sweden).

The bacteria-expressed his-fused S proteins were subjected to sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS–PAGE). We transferred the proteins onto nitrocellulose membrane (Amersham Biosciences, UK), which was then blocked with 5% skim milk in TBST (TBS with 0.05% Tween 20) for 1 h. Polyclonal mouse anti-his antibodies (Amersham Biosciences, UK) or convalescent sera of SARS patients were added and incubated for additional 1 h. The membrane was washed with TBST three times at 5 min/each. We detected the bound antibodies by adding horseradish peroxidase (HRP)-conjugated anti-mouse antibodies or anti-human Ig antibodies (Sigma, USA). After three washings, we developed the membrane with diaminobenzidine (DAB) substrate until the desired intensity.

Female White Leghorn (Gallus domesticus) chickens were immunized with mixed recombinant S proteins (10 μg/each) in equal volume of adjuvant by intra-muscular injection. The first immunization was prepared with complete Freund's adjuvant, and three additional ones were performed with incomplete Freund's adjuvant at interval of 7 days. After each immunization, we obtained the blood and titrated it with enzyme-linked immunosorbent assay (ELISA) to determine the presence of an antigen-specific immune response. We also collected eggs for purification of immunoglobulin Y (IgY) from yolks by the standard procedure (Akita and Nakai, 1993a, Akita and Nakai, 1993b).

We constructed the library according to the published protocols with minor modifications (Andris-Widhopf et al., 2000). Briefly, 7 days following the final immunization, chickens were euthanized and spleens were harvested. Spleens were placed immediately in Trizol reagent for homogenization. The extraction of total RNAs was carried out and 10 μg of RNAs was reversely transcribed to synthesize the first-strand cDNA using a SuperScript RT kit (Invitrogen). After PCR amplification with chicken specific primers, we obtained gene products of heavy and light chain variable regions (V_H and V_L). For full-length single chain (sc) Fv construction, we connected V_H and V_L fragments with a peptide linker (GGSSRSS) by an overlap extension PCR, which was further cloned into a pComb3H vector. The recombinant DNAs were transformed into E. coli XL-1 strain by electroporation. Recombinant phage production was initiated by the addition of helper phage VCS-M13 (Barbas et al., 1991), precipitated with 4% polyethylglycol 8000 and 3% NaCl (w/v), resuspended in phosphate-buffered saline (PBS) containing 1% bovine serum albumin (BSA) and stored at 4 °C.

We coated microtitre plates with equal amounts of the mixed S-protein fragments (0.5 μg/well) at 4 °C overnight. After being blocked with 3% BSA at 37 °C for 1 h and washed with TBST, we added more than 10¹¹  pfu of recombinant phages from above to each well and the plates were incubated at 37 °C for 2 h; unbound phages were removed and the wells were washed vigorously with TBST buffer. Bound phages were eluted with 0.1 M HCl/glycine (pH 2.2)/0.1% BSA, and neutralized with Tris base buffer. Eluted phages were used to infect E. coli XL-1 strain at log phase for amplification and rescued with 4% polyethylglycol 8000 and 3% NaCl for the next round of selection. Phage preparation and further panning on wells were repeated four times. For preparing single phage clone, individual colonies from the final panning cycle were grown in 10 ml SB medium containing ampicillin (100 μg/ml) at 37 °C with agitation for 4–6 h. VCS-M13 helper phage was added to the cultures and shaken for another 2 h. Kanamycin was added to a final concentration of 70 μg/ml. Individual phage clones were rescued, as above, and tested for their binding ability.

We used microtitre plates precoated with SARS-CoV-infected Vero E6 cell lysates to detect the binding ability of scFv-expressing phage to S protein, as shown in the anti-SARS-CoV ELISA kit (Euroimmun, Lubeck, Germany). After rinsing with TBST in the plates, we distributed 10¹⁰ phage particles dissolved in PBS to wells in duplicate and incubated them at 37 °C for 1 h. After washing with PBST (PBS with 0.05% Tween 20), we detected bound phage particles with HRP-labeled mouse anti-M13 antibody (Amersham Biosciences). In the epitopic mapping assay, we coated 10 purified his-fused S-protein fragments in microtitre plates at 0.5 μg/well individually. After being blocked with 3% BSA, we distributed 10¹⁰ phage particles to each well in duplicate and incubated them at 37 °C for 1 h. After washing, the detection was processed as described above.

Biochip slides (Euroimmun, Lubeck, Germany) were used to test the binding reactivity of antibodies in sera or scFv-expressing phage to S protein in SARS-CoV-infected Vero E6 cells. After rinsed with TBST, slides were incubated with either 10¹¹ phage particles, immunized chicken sera or human sera for 1 h at RT. After washed with TBST, the bound antibodies or phage were detected with mouse anti-M13 antibodies or corresponding secondary antibodies (Sigma, USA) conjugated with fluorescein isothiocyanate (FITC). Slides were coated with mounting oil and inspected with an immunofluorescent microscope.

The nucleotide sequences of V_H and V_L genes from selected clones were determined by an auto-sequencer (ABI PRISM 377; Perkin-Elmer) using ompseq (5′-AAGACAGCTATCGCGATTGCAGTG-3′) and HRML-F (5′-GGTGGT-TCCTCTAGATCTTCC-3′) primers.

To produce antigens for animal immunization and epitopic mapping, 10 fragments of S protein gene were PCR-amplified, cloned into pET-21 vector and expressed as his-fused recombinant proteins. The locations and predicted gene lengths of these fragments are shown in Fig. 1A and B. After induction, these fragments were successfully expressed in E. coli BL-21 cells with molecular weights of 12–30 kDa. Individual fusion protein was purified by Ni²⁺-charged sepharose and analyzed on SDS–PAGE (Fig. 2A). In Fig. 2B, anti-his antibodies were used to further confirm their identity on Western blot.Fig. 1

(A) Schematic drawing of cloning strategy for 10 truncated S genes. (B) Amino acid residual locations and predicted molecular weights of these truncated proteins expressed in E. coli cells.

Western blot analysis was used to determine antigenic reactivity of 10 SARS-CoV-infected patients’ sera and three normal control sera. As shown in Table 1 , fragment Sg-g had apparent binding reactivity with all patients’ sera, while three other fragments, Sb-b, Se-e and Sf-f, were recognized by 70% of patients’ sera. Only 20% of patients’ sera recognized fragments Sc-c and Sh-g. Four remaining fragments showed no binding reactivity to all patients’ sera as well as to three normal sera. The representative Western blots of these results are shown in Fig. 3A (reacted with one patient's serum) and Fig. 3B (with one normal serum). On contrast to normal sera, most patients’ sera could clearly recognize the fragments of Sb-b, Se-e, Sf-f and Sg-g. The results suggested that fragment Sg-g alone or together with the three other fragments may serve as a good target to detect anti-SARS-CoV S protein antibodies. To generate S protein-specific phage clone, 10 S fragments (15 μg/fragment) were mixed to immunize chicken from which spleens had been taken for scFv antibody library construction.Table 1

Reactivity of his-fused S proteins with patient and normal sera

The amplification of scFv immunoglobulin genes in chicken was performed by two consecutive PCR steps to produce final products with bands of the expected 750 base pair (bp). In the primary PCR, V_H and V_L gene were amplified as 400 and 350 bp in size, respectively. Difference in intensity was observed, which may reflect variation in the gene expression and biased amplification efficiency intrinsic to each chicken (data not shown). Thereafter, a library (4S) containing about 5 × 10⁷ clones was constructed. Restriction analysis of 10 randomly chosen clones from the library indicated that 8/10 (80%) of clones contained scFv gene inserts (data not shown). The SARS-CoV S protein-specific binders were enriched by four rounds of panning. Then, individual phage clones were amplified and analyzed for their binding activity by phage ELISA. The results in Fig. 4 showed that the randomly chosen phage clones exhibited significant binding activity (OD between 0.7 and 1.2) and six clones showed intermediate level of binding activity (OD between 0.4 and 0.6) against SARS-CoV-infected Vero E6 cell lysates. In control samples, both wild-type M13 phage and irrelevant scFv-expressing phage showed weak reactivity (OD below 0.2). Accordingly, three highly and two moderately positive clones, including 4S1, 4S3, 4S4, 4S12 and 4S15, were further characterized. To understand the scFv gene diversity of these clones, V_L and V_H genes were sequenced by two designed sense primers (Barbas et al., 2001). After aligning with the chicken germline gene, the five chosen clones revealed 6–12% and 31–62% variation in framework (FR) and CDR regions, respectively, leading to a 13–25% mutation rate in the entire V region (Fig. 5 and Table 2 ).Fig. 4

Binding activity of randomly chosen phages from library 4S after the fourth panning cycle. Fifteen phage clones (4S1 to 4S15) were amplified and prepared to test their binding to SARS-CoV-infected Vero E6 cell lysates in an ELISA test. Wild-type M13 phage and one irrelevant phage expressing anti-Eno1 scFv were used as negative controls. Binding phages were detected by anti-M13 phage antibodies conjugated with horseradish peroxidase. The color was developed and absorbance was measured at 450–650 nm.

We used immunocytochemical staining to determine binding ability of scFv-expressing phage 4S1 to the SARS-CoV-infected Vero E6 cells (Fig. 6 ). Four other phage clones, positive in the ELISA method, were also examined; however, they only had weak reactions (data not shown). Bound phages were detected with FITC-labeled anti-M13 antibodies. Patient and normal human sera were used as positive and negative controls. Sera from immunized and pre-immunized chickens were also included. Staining data clearly showed that both polyclonal IgY antibodies (Fig. 6C) and phage 4S1 (Fig. 6E) bind to SARS-CoV-infected Vero E6 cells, whose binding activity is comparable to that of convalescent patient serum (Fig. 6A). Nevertheless, all negative controls, including normal human serum (Fig. 6B), pre-immunized serum (Fig. 6D) and wild M13 phage (Fig. 6F), exhibited no or little binding signal. These results suggested that the bacteria-derived S protein fragments were able to mimic the conformation of native S protein and induced the humoral immune response in chicken. Moreover, specific scFv antibody on phages could be generated to recognize the native S protein expressed on the SARS-CoV-infected Vero E6 cells.Fig. 6

Binding activity of 4S1 phage binder to native S protein. Commercially available Biochips, precoated with SARS-CoV-infected Vero E6 cells were immunostained, respectively, with sera from a convalescent-stage patient (A), un-infected person (B), post- (C) and pre-immunized (D) chickens, 4S1 phage (E) and wild-type M13 phage (F).

The 4S1 phage binder was used to define the possible antigenic site on the S protein. After being reacted with individual truncated S protein fragments, 4S1 phage showed distinct binding activity against Se-e fragment, suggesting a dominant epitope existed in this location (Fig. 7A). On the contrary, an irrelevant phage clone expressing anti-ENO1 scFv showed no apparent binding activity (OD below 0.2) in the negative control test (Fig. 7B).Fig. 7

Epitope mapped by 4S1 phage binder. 4S1 phage (A) and an anti-ENO1 phage (B) were examined for their binding activity to 10 purified and truncated S protein fragments coated on ELISA plates.