Neutralization epitope responsible for the hepatitis B virus subtype-specific protection in chimpanzees Pei Zhang*†, Mei-ying W. Yu*, Richard Venable‡, Harvey J. Alter§¶, and J. Wai-Kou Shih§储 *Division of Hematology and ‡Laboratory of Biophysics, Center for Biologics Evaluation and Research, Food and Drug Administration, Bethesda, MD 20892; and §Department of Transfusion Medicine, Warren Grant Magnuson Clinical Center, National Institutes of Health, Bethesda, MD 20892 Contributed by Harvey J. Alter, May 9, 2006

Neutralizing monoclonal antibody (BX-182) directed against the d determinant of hepatitis B virus (HBV) surface antigen protected chimpanzees from infection by HBV subtype adw but not by subtype ayw, as demonstrated by intravenously inoculating a mixture of the antibody with the respective subtype of the virus. To elucidate the mechanism underlying the subtype-specific protection, a combinatorial approach of screening random peptide phage libraries, bioinformatics, and structure analysis was used in this study to identify the neutralization epitope responsible for the observed protection. The epitope was mapped at the N terminus of the pre-S1 region of the hepatitis B surface antigen between residues 17 and 21, of which the residues Val-18兾Pro-19 were critical for antibody binding. Alignment of amino acid sequences derived from diverse genetic variants of HBV revealed that the epitope was present in ad subtypes and in their corresponding genotypes A, B, C, F, and H. By contrast, this epitope was not found in a majority of ay subtypes or in genotypes D, E, and G, where the antigenic residues Val-18兾Pro-19 within the epitope were replaced by Thr兾Ser, Thr兾Thr, or Ala兾Ser, respectively, resulting in a drastic conformational change of the epitope. These data indicate that, by binding discriminately to the subtype ‘‘d’’ epitope in the pre-S1 region, neutralizing antibody BX-182 protects chimpanzees from HBV infection in a subtype-specific manner, suggesting a potential escape mechanism for HBV genetic variants. hepatitis B surface antigen 兩 neutralizing antibody

H

epatitis B virus (HBV) is a blood-borne hepatotrophic virus that infects an estimated 350 million people worldwide (1). Besides the manifestations associated with acute hepatitis, chronic HBV infection constitutes a significantly high risk for the development of liver cirrhosis and hepatocellular carcinoma (2). HBV is a small enveloped partially double-stranded DNA virus that belongs to the family Hepadnaviridae. The HBV surface antigen (HBsAg), a specific serologic marker, is composed of the large (L), middle (M), and small (S) protein subunits. These subunits are encoded by a single ORF containing three in-frame translational initiation codons and share a common C-terminal S ORF, which defines the pre-S1 (residues 1–108 or 1–119, depending on the serological subtype), pre-S2 (residues 120–174), and S (residues 175–400) regions. The S protein serves as a membrane anchor and plays an important role in virus assembly and, possibly, membrane fusion. The M protein contains the pre-S2 and the S regions. The L protein contains all three regions; it is preferentially present on the infectious virus particle and is essential for both viral assembly and infectivity (3). HBsAg derived from different strains carries serologically defined group-specific determinants, designated by a common a determinant and two sets of mutually exclusive subdeterminants d兾y and w兾r. Therefore, four HBsAg subtypes, including adw, adr, ayw, and ayr, represent the major viral phenotypes (4). Recently, according to the homogeneity of the virus sequence, eight HBV genotypes, A–H, have been classified based primarily on an intergenotype divergence of ⬎8% (5–8). The correlation between serologic subtypes and genotypes has been partially established (9). Strikingly, the prevalence of different genotypes varies geographically and is strongly associated with 9214 –9219 兩 PNAS 兩 June 13, 2006 兩 vol. 103 兩 no. 24

ethnicity. Genotype A is prevalent mainly in northwestern Europe and North America (10). Genotypes B and C are highly prevalent in Asia. Genotype D has been distributed worldwide, but is predominant in the Mediterranean region. Genotype E is restricted almost entirely to West Africa, and genotype F is prevalent in Central and South America. Genotype G was found in Europe and the United States (8). However, intertypic recombinations between different HBV genotypes have been noted recently (11–14). Studies relating to the impact of viral variation on the clinical course of disease have drawn considerable recent attention (10, 15). HBV vaccine and plasma-derived hepatitis B immune globulin (HBIG) have been very successful in both actively and passively neutralizing HBV infection. However, the emergence of HBV variants兾mutants in vaccinated children (HBIG and HBV vaccine) and the limited availability of plasma-derived HBIG are growing concerns (16–18). Unfortunately, because of the species-specificity of HBV infection, the efficacy of neutralizing antibodies in vivo requires evaluation in chimpanzees, although in vitro models have shown some promise (19, 20). Despite this limitation, increasing evidence demonstrates that humoral immunity is important for protection from HBV infection. Earlier studies demonstrated that antibodies against the common a determinant of the S protein or the pre-S1 peptide (residues 21–47) neutralized HBV infection in chimpanzees and humans (21–28). Recent studies, using primary hepatocytes from Tupaia belangeri as a model system, mapped an essential domain for the virus binding to the receptor at the N terminus of pre-S1 (19, 29). In this study, using a combinatorial approach of screening random peptide phage display libraries, bioinformatics and analysis of structure as a function of sequence, we have identified a neutralization epitope responsible for an antibody exerting its subtype-specific protection in chimpanzees. This study illustrates a molecular mechanism for the neutralization of HBV infection in a subtype兾genotype-specific manner. Results Monoclonal Antibody BX-182 Preferentially Recognizes the adRelated Subtypes of HBsAg. The subtype specificity of BX-182 was

determined in a modified Ausab competition assay (Abbott) by using a panel containing seven HBsAg-positive sera (4). As a result, as illustrated in Table 1, BX-182 was strongly reactive to the ad-related subtypes in the test panel, including adw and adw4, and partially reactive to the adr subtype. By contrast, BX-182 did Conflict of interest statement: No conflicts declared. Freely available online through the PNAS open access option. Abbreviations: ALT, alanine aminotransferase; CID, chimpanzee-infectious dose; HBV, hepatitis B virus; HBsAg, HBV surface antigen; RMSD, rms difference. †To

whom correspondence may be addressed at: Division of Hematology, Center for Biologics Evaluation and Research, Food and Drug Administration HFM-345, Room 305, 29 Lincoln Drive, Bethesda, MD 20892. E-mail: [email protected].

¶To

whom correspondence may be addressed at: Department of Transfusion Medicine, Warren Grant Magnuson Clinical Center, National Institutes of Health, Bethesda, MD 20892. E-mail: [email protected].

储To

whom correspondence may be sent at the present address: National Institute of Diagnostic and Vaccine Development, Xiamen University, Xiamen 361005, China. E-mail: [email protected].

www.pnas.org兾cgi兾doi兾10.1073兾pnas.0603316103

Table 2. Reactivity of BX-182 with phage clones

Inhibitory activity, % Subtype adw ayw-4 adr ayw-1 ayw-2 ayw-3 ayr

BX-182 100 97 42.1 7.5 11.4 3.2 12

Rb8202 100 0.7 0 93.8 102 0 15

not show any significant binding to ay-related subtypes. As a control, Rb8202, a rabbit antiserum known to react with the common a determinant of HBsAg, showed no preference between ad- and ay-related subtypes, although it failed to bind some subtypes, including adw4, adr, ayw, and ayr. One of the simplest explanations is that Rb-8202 reacts with only a part of ‘‘a’’ determinant. Nevertheless, these data demonstrated that BX-182 could discriminate ad subtypes from ay subtypes. Subtype-Dependent Protection of Chimpanzee from HBV Infection by BX-182. To test whether BX-182 could neutralize HBV infectivity

in vivo, CH-1404 was inoculated with the mixture of monoclonal antibody BX-182 and an HBV inoculum of adw subtype at 103 chimpanzee-infective dose (CID)50. As depicted in Fig. 1, BX182 neutralized the infectivity of HBV inoculum of adw subtype in the chimpanzee. CH-1404 exhibited neither elevated alanine aminotransferase (ALT) nor detectable HBsAg or anti-HBc over a period of 44 weeks. To determine whether CH-1404 had remained fully susceptible to infection with HBV, it was challenged again, in the absence of BX-182, at week 55 by the same inoculum containing 103 CID50 of HBV adw subtype. As expected, both HBsAg and anti-HBc became positive at week 16 after challenge, and serum

Fig. 1. Subtype-specific protection of chimpanzees from HBV infection by BX-182. HBV antigen and antibody in serum were scored as positive when the signal-to-noise (S兾N) was ⱖ2.1 by radioimmunoassay. ALT was regard as elevated when it reached the level twice the upper limit of normal. EIA, enzyme immunoassay.

Zhang et al.

Library

Phage clones sequenced

C7C

10

M12

11

Phage-displayed peptide TNPVLRS (9) LPGTPHI (1) SVPPPHTRSASG (6) MEGQYKSNLLFT (5)

Numbers of peptides found in DNA sequencing are in parentheses.

ALT levels became elevated at week 15, thereby demonstrating susceptibility of the chimpanzee to HBV infection. These data demonstrated that BX-182 neutralized the infectivity of the HBV adw subtype in the chimpanzee model. In contrast, chimpanzee CH-1419, infused with an incubation mixture of the inoculum of ayw subtype and BX-182 antibody, was infected by HBV and developed hepatitis B. Serum HBsAg became positive at week 6 after inoculation and remained positive for 20 weeks. Seroconversion, as indicated by the appearance of anti-HBc, was observed 3 weeks after inoculation. The ALT level was elevated at week 14. These results indicate that BX-182 did not react with HBV ay subtypes in vitro and demonstrate that BX-182 blocked HBV infection in vivo in a subtype-dependent manner. Mapping of Neutralization Epitope. Blocking the virus infectivity in

a chimpanzee by the binding of BX-182 antibody to HBV adw inoculum prompted us to map the neutralization epitope(s). Initially, we screened a random peptide phage library (C7C) with the BX-182 antibody. Because the phage displayed a looped heptapeptide, we reasoned that the local ‘‘criticality’’ of nonlinear amino acid residues could be revealed by their recognition by BX-182 antibody. As indicated in Table 2, after four rounds of selection, 9 of 10 sequenced clones had the same displayed peptide (ThrAsnProValLeuArgSer). These data suggested that that these amino acid residues, irrespective of their linear organization, mimicked the conformation of the BX-182-binding site on HBsAg. Once the critical amino acid residues within the epitope were identified, we screened a second library (M12) displaying a linear dodecapeptide to further determine the linear arrangement of these residues. As a result, two relevant peptide sequences were identified after three rounds of selection. The first peptide, SerValProProProHisThrArgSerAlaSerGly, showed a preferential usage of residues that were similar to the peptide ThrAsnProValLeuArgSer, detected by phage library C7C. In particular, both Val and Pro residues and, to some extent, the Ser residue, were found to be essential components for antibody binding. The second peptide showed no significant similarity to the looped heptapeptide or obvious sequence homology to the HBsAg. Whether this peptide mimicked the conformation of the antibody-binding site requires further investigation. To relate phage-displayed peptides to the antibody-binding site(s) on the HBV, we performed BLAST searches on the HBsAg sequence of adw subtype. It was evident that the N terminus of the linear dodecapeptide SerValProProProHisThrArgSerAlaSerGly had a significant sequence homology with the pre-S1 region of HBsAg (Fig. 2). Notably, the wild-type adw subtype sequence contained SerValProAsnPro. The replacement of Asn by Pro in phage-displayed peptides suggested that Asn was not crucial for the binding. Knowing that the looped heptapeptide (ThrAsnProValLeuArgSer) mimicked the pre-S1 epitope by a spatial arrangement of four residues (Asn, Pro, Val, and Ser), this study demonstrated that the neutralization epitope was located at the N terminus of the pre-S1 region of HBsAg from HBV adw subtype, precisely between residues 17 and 21. PNAS 兩 June 13, 2006 兩 vol. 103 兩 no. 24 兩 9215

MEDICAL SCIENCES

Table 1. Subtype specificity of BX-182

Fig. 2. Mapping of pre-S1 amino acid sequence required for BX-182 binding. The peptides identified from each phage library, either linear or conformational, are indicated. Residues that are identical to the pre-S1 region (adw) are underlined by an arrow. The arrangement of the residues in the peptide critical for the binding of BX-182 is indicated by the direction of the arrow.

Mechanism for Subtype-Dependent Neutralization. Because the pro-

tection by BX-182 was subtype-specific, we examined whether subtype-dependent neutralization could be attributed to the differences in sequence of identified epitopes among subtypes (Fig. 3). We found that the identified epitope was present in ad-related subtypes but absent in a majority of ay-related subtypes. Furthermore, depending on ay subtypes, the two critical residues Val-18兾Pro-19 within the epitope were replaced by residues of Thr兾Ser and Thr兾Thr. Therefore, residues Val兾 Pro inside the neutralization epitope constituted a core sequence essential for antibody binding. To determine the occurrence of the neutralization epitope in various HBV genotypes, we analyzed sequences of all known HBV genotypes (A–H) for epitope variants (Fig. 3). The identified epitope was found in genotypes A, B, C, F, and H and was absent from genotypes D, E, and G, where Val-18兾Pro-19 were, again, replaced by residues of Thr兾Ser, Thr兾Thr, and Ala兾Ser, respectively. Thus, these data established a clear correlation between the subtype and genotype with regard to the occurrence of this functional neutralization epitope. Structural Basis of Neutralization Epitope and Its Variants for the Binding of BX-182. Elucidation of epitope structure is an integral

part of studying antigen–antibody interactions. However, lack of the crystal structure of HBsAg hampered our effort to precisely demonstrate the conformational differences between the neutralization epitope and its variants. Consequently, the structural tendencies of the pentapeptides for the epitope and variants were studied by comparative analysis of crystals with these short

Fig. 3. Alignment of amino acid sequences of the pre-S1 regions among various HBV genotypes. A consensus sequence of N-terminal pre-S1 between amino acid residues 1 and 40 is shown. The residues of genotypes A–H contributing to the subtype-specific protection are shaded. A hyphen indicates an amino acid residue identical to that of the consensus sequence. An empty space represents an amino acid deletion.

sequences and by theoretical calculations using a simulated annealing protocol. Sequence searches of protein crystal structures deposited in the Protein Data Bank located only one protein with the wild-type SVPNP sequence (glutaminyl tRNA synthetase), so the conservative point mutants TVPNP and SVPQP were included to broaden the sample. A comparison of structural features of representative proteins for both these and the nonantigenic variants STSNP, STTNP, and SASNP is presented in Table 3; backbone aligned pentapeptides from these proteins are also depicted in Fig. 4. The most significant point is the greater conservation of backbone structure for the VPcontaining sequences, which tends to present the hydrophobic Val and Pro side chains in a fixed orientation with respect to each other. The secondary structure classification and dihedrals listed in Table 3 suggest that the xVPxP sequences have similar structures (Fig. 4). On the other hand, a greater variability and different turn shapes were observed for the non-VP variants. The conformational tendencies were further studied by analyzing the results of simulated annealing calculations for five sequences: SVPNP, STTNP, STSNP, SASNP, and AAAAA as a control. The cluster analysis, based on the backbone flexibility, showed a small number of discrete conformational substates found for the SVPNP sequence, indicating a greater rigidity (Table 4). The average pair-wise rms difference (RMSD), indicated as ⬍RMSDij⬎ and its standard deviation, are also both significantly smaller for the SVPNP epitope sequence, further indicating greater rigidity than any of other epitope variants.

Table 3. Protein crystal structural analysis of the SVPNP epitope, conservative point mutants, and nonantigenic variants Sequence

PDB

Protein

SS Class*

␾2

␺2

␾3

Replicas†

⫺120

155

⫺66

⫺99

130

⫺53

1EUQ 1EUY 1EXD 1GTS 1NYL 1O0B 1O0C 1QRS 1QRT 1QRU 1QTQ 1ZJW 1I24 1I2C 1QRR

Coil Turn IV Turn I

⫺127 ⫺104 ⫺85

164 ⫺3 ⫺41

⫺64 ⫺167 ⫺107

1ZP9(3) 1ZTH(4)

Turn IV Turn IV Turn IV Turn I

⫺153 ⫺118 ⫺134 ⫺43

148 174 140 ⫺59

⫺92 ⫺118 ⫺61 ⫺113

1DYS(1)

Turn IV Turn I

⫺62 ⫺80

140 ⫺14

⫺89 ⫺112

SVPNP

1GTR

Glutaminyl tRNA synthetase

Turn VIII

TVPNP

1I2B

Coil

SVPQP STSNP

1ZP9 1AYX 1H80

Sulfolipid biosynthesis protein SQD1 Serine kinase Glucoamylase Galactohydrolase iota-carrageenase Endoglucanase T cell surface glycoprotein CD4 AMP nucleosidase Human rhinovirus 14 coat protein Benzoylformate decarboxylase MLN64 lipid-transport domain

STTNP SASNP

1DYS 1CID 1T8Y 1RVF 1Q6Z 1EM2

1H80(1) 1KTW(2)

1TY8(5) 1NCQ

The portions of the sequences in bold are significant. *Secondary structure (SS) classification and torsion values determined by using the STRIDE program. †Replicas are subunits of a multimer, duplicate PDB depositions of the same protein, or both; number in parentheses indicates subunits. 9216 兩 www.pnas.org兾cgi兾doi兾10.1073兾pnas.0603316103

Zhang et al.

Table 4. Simulated annealing calculation of the SVPNP epitope, conservative point mutation, and nonantigenic variants Number of clusters at radius R

25 30 35 40 45 50 具RMSDij典 Average SD

Fig. 4. Backbone-aligned representative pentapeptides from Protein Data Bank crystal structures. (Top) The SVPNP epitope (1GTR) and conservative mutants TVPNP (1I2B) and SVPQP (1ZP9); note the conserved placement of Val and Pro residues. (Middle) The STSNP (1AYX, 1H80, 1DYS, and 1CID) and STTNP (1T8Y) mutants are shown; note the greater variation in backbone and side-chain positions and the absence of a conserved hydrophobic patch at positions 2 and 3. (Bottom) The SASNP (1RVF, 1Q6Z, and 1EM2) mutant; in addition to the differences noted for STSNP and STTNP, the turn tends to be sharper as well.

Discussion Illustrating the mechanism for HBV subtype兾genotype-specific neutralization is of great importance with regard to vaccination and therapy with hyperimmune globulin. Earlier experiments showed that cross-protection could be achieved when chimpanzees were vaccinated with S protein of HBsAg, irrespective of the subtype used in the vaccine or challenge, consistent with the conserved nature of the common a determinant. However, the majority of antibodies generated immediately after vaccination were subtypespecific. For instance, immunization of humans or chimpanzees with S protein of adw initially gave rise to d-specific antibodies, and only with time was the immune response broadened to include antibodies against the a determinant (30, 31). Subtype specificity was further demonstrated by the influence of viral subtypes on both humoral and T cell immune responses to the pre-S2 (32, 33). However, demonstration of the functional significance of these subtypes has been difficult because there is no currently available assay that can discriminate subtle changes in amino acid sequence and precisely predict the corresponding function. In this study, we showed that monoclonal antibody BX-182 is highly, if not exclusively, d subtype-specific in vitro. More importantly, BX-182 is able to block HBV infectivity of the adw, but not the ayw, subtype in chimpanzees. Zhang et al.

SVPNP 5 3 2 2 2 2 1.13 0.62

STTNP

STSNP

SASNP

12 16 18 7 13 12 6 95 12 6 7 11 5 6 8 5 5 7 Backbone RMSD for all possible pairs 1.42 1.62 1.80 0.76 0.79 0.85

AAAAA 215 16 16 13 12 11 1.98 0.94

It was this unique feature of BX-182 that allowed us to apply random peptide phage-display techniques to map the HBV neutralization epitope of BX-182. The neutralization epitope was found to be located at the N terminus of the pre-S1, precisely between amino acid residues 17 and 21. This result was unexpected, because it had been reported that the d兾y subtypes were determined by residue 122 of the S antigen (7, 34). Nevertheless, our results suggest that the distinctions between d and y subtypes may be determined by at least two epitopes. The mechanism for the subtype-specific protection was illustrated in our study. By comparing the amino acid sequences of the adw and ayw subtypes, we found that subtype specificity was determined by the binding of neutralizing antibody to a region of epitope where the subtypes differed by two amino acid residues. The epitope in adw contains 18Val兾19Pro, whereas these amino acids are replaced by hydrophilic residues Thr兾Ser in the ayw1, 2, and 3 subtypes. As a consequence of these substitutions, the conformation of the epitope, as predicted by 3D modeling and analysis of crystal structures, was drastically changed. Our data, therefore, provide a molecular basis for the differential HBV subtype ad兾ay protection observed in chimpanzees. The location of the neutralization epitope at the N terminus of pre-S1 has virologic relevance because pre-S1 had been suggested to be important for the attachment and uptake of HBV. In fact, peptide derived from pre-S1 between residues 21 and 47 of genotype A could inhibit the binding of HBsAg to both human hepatic cells (35) and human hepatic membranes (36), providing a basis for the design of a pre-S1 (residues 2–36)-containing hepatitis B vaccine that has been licensed in Europe (37). Supporting these studies, recent experiments using the primary tupaia hepatocyte culture as a model demonstrated that HBV infectivity could be inhibited efficiently by pre-S1 lipopeptides and suggested that the essential domain for the virus binding was located between residues 20 and 29 (29, 38). Our data identify a neutralization epitope that partially overlaps the essential virus-binding domain described in these studies. Although we have not proven that the identified epitope is directly involved in the binding of HBV to its hepatocyte receptor, the steric hindrance generated by the binding of BX-182 to the neutralization epitope contributes to the disruption of the interactions of HBV with hepatocytes in vivo. The data presented here are relevant to our understanding of the interplay between HBV and host immune responses. Currently, there are at least eight different genotypes and four major subtypes of HBV. In chronic carriers, these naturally occurring variants may exist as the predominant HBV population as a result of selective pressures. By demonstrating that not all subtypes兾genotypes possess the pre-S1 neutralization epitope, we suggest that variants lacking neutralization epitopes may have advantages in escaping antibody response. This concept needs to be developed further because, in a clinical setting, depending on the harbored virus genotypes兾subtypes in patients, the initial sensitivity of virus to a PNAS 兩 June 13, 2006 兩 vol. 103 兩 no. 24 兩 9217

MEDICAL SCIENCES

R

neutralizing antibody dictates clinical outcomes (15). Therefore, for a successful prophylactic and therapeutic intervention, humoral immune responses must be exceptionally broad and potent against multiple neutralization epitopes to control preexisting variants and to limit viral evolution. Our findings, thus, open up a new possibility to improve prophylaxis and treatment of HBV infection as well as HBV vaccine design based on subtype-specific epitope structures. Materials and Methods Antibodies. A monoclonal antibody, BX-182, was raised against HBsAg after immunization of mice with HBsAg of adw subtype (39). The specificity for the subdeterminants of HBsAg, as demonstrated by a radioimmunoprecipitation assay, was identified by its reaction with [125I]HBsAg of ad, but not ay, subtype. Before this antibody was used, it was further purified by protein A-affinity chromatography. As an experimental control, rabbit anti-S1-20 was generated by immunizing with a synthetic peptide composed of residues 1 and 20 of the N terminus of S protein. HBV Inocula. Two standard wild-type HBV inocula, adw and ayw,

kindly provided by Robert Purcell (National Institutes of Health), were used for this experiment. The infectivity of these strains in chimpanzees had already been determined. Chimpanzee Studies. Two chimpanzees, CH-1404 and CH-1419, were used in this study. The housing, maintenance, and care of these chimpanzees met or exceeded all relevant guidelines and requirements. These chimpanzees had not been inoculated with any HBV-containing materials and were seronegative for all HBVassociated serologic markers. CH-1404 was given an i.v. injection of a mixture of monoclonal antibody BX-182 with an HBV inoculum of adw subtype at 103 CID50. The antibody was present in a 740-fold excess over equivalent antibody兾antigen ratios, as determined by an antibody-inhibition assay. CH-1419 was infused with an HBV inoculum of ayw subtype at 103 CID50 that was preincubated with the same quantity of anti-adw monoclonal antibody BX-182. Each mixture was prepared by incubation at 4°C overnight before inoculation. After i.v. infusion of the antibody and virus mixture, serum samples were collected weekly. Hepatitis B Markers. Serum ALT levels were measured by standard methods (Antech, Baltimore). ALT was considered elevated when it reached levels twice the upper limit of normal. Sera were also assayed weekly with an enzyme immunoassay method (Abbott) for serological markers of hepatitis B, including HBsAg, anti-HBs, and anti-HBc. Phage Display. Two random peptide phage-display libraries, C7C and M12, were used for mapping the HBV epitopes (40). The C7C library displayed seven amino acid residues by flanking the two designed cysteine residues on the coat protein pIII of M13 filamentous phage. The two Cys residues formed a disulfide bond to present a loop structure, permitting the local criticality of nonlinear amino acid residue arrangements. M12, on the other hand, was a linear peptide library. Each phage clone displayed a dodecapeptide on the surface of the pIII of M13. These two libraries contained at least 109 independent clones. Selection Phage Clones. The two phage-displayed random peptide

libraries were biopanned independently with the monoclonal antibody BX-182; a protein A-immunoprecipitation method was used to capture reactive phage clones. Briefly, phage (109 plaque-forming units) were incubated with BX-182 in Trisbuffered saline (pH 7.2) containing 0.1% Tween 20 for 20 min at room temperature, followed by adding protein A beads for an additional 20 min. Unbound phage were removed by a series of 10 2-min washes with Tris-buffered saline (pH 7.2) containing 0.2% Tween 20. Captured phage were eluted by adding 100 ␮l 9218 兩 www.pnas.org兾cgi兾doi兾10.1073兾pnas.0603316103

of 0.1 M HCl and shaking vigorously for 10 min. The eluted phage clones were immediately neutralized with 33 ␮l of 1 M Tris䡠HCl (pH 8.0). Half of the eluted phage were amplified in Escherichia coli ER 2738 and harvested for further biopanning. After three or four rounds of selection and amplification, the specific region of phage DNA, which was extracted from the individual clones, was sequenced. Peptide sequences were deduced by GENERUNNER software from phage DNA sequence. Bioinformatics Analysis. DNA sequences derived from different

genetic variants were retrieved from the National Center for Biotechnology Information database. We performed BLAST searches with the peptide sequences selected by the monoclonal antibody BX-182 to identify homology with HBsAg L protein. The algorithm was based on the alignment of a pair of linked amino acid residues within the antigen to generate the core residues of the predicted epitope. Based on the location of core residues within the L protein of HBsAg and the frequency of selected amino acid residues within the reactive phage peptide from both libraries, the validity of the interpretation of epitope was made. Peptide Structural Analysis. The relationship between pentapeptide sequence and 3D structure was examined by two distinct approaches. First, the RCSB Protein Data Bank of macromolecular structure data (41) was searched for the SVPNP epitope sequence and conservative mutants TVPNP and SVPQP and for the variant sequences with VP replaced by TS, TT, or AS. Each of the proteins found was analyzed by using the STRIDE program (42) for secondary structure classification based on the ␾,␺ dihedral angles for the pentapeptide sequences. In the first phase, replicate structures were compared, such as multiple depositions of the same protein and proteins with multiple identical subunits. For all such cases, the variability in the backbone torsions was small, within roughly 5–10° of each other, and a single protein or subunit was chosen to represent all of the replicas. In the second approach, the sequences SVPNP, STSNP, STTNP, SASNP, and AAAAA were subjected to a simulated annealing procedure (43) to generate 200 structures for each sequence, which were then analyzed for structural tendencies; the sampling and analyses were performed by using the CHARMM program (44). In brief, a pentapeptide with neutral end caps (acetyl and methylamine) for each sequence was rapidly heated to 600 K and cooled in stages to 100 K by using low-friction Langevin dynamics; the procedure was run 200 times for each sequence, with different initial velocities. Each of the ensembles of 200 structures was analyzed by clustering of the backbone ␾,␺ torsions, and by a full-backbone atom RMSD pair-wise cross-analysis (N, C␣, and C atoms only). In the cluster analysis, the number of clusters found as a function of the cluster radius R was evaluated as a measure of structural diversity. The clustering uses a straightforward leastsquares approach; a given conformation belongs to a cluster when

冑冘

(␪k ⫺ ␪°k)2兾n ⬍ R,

[1]

k ⫽ 1,n

where ␪k is one of the backbone torsions, and ␪k° is the average ␪k for current members of the cluster. The clustering method accounts for periodicity, i.e., 175° and ⫺175° are only 10° apart and not 350° apart when computing the difference. The other main analysis of the simulating annealing results was based on a full pair-wise RMSD of the backbone atoms for all 200 structures after alignment. The statistic ⬍RMSDij⬎ was computed as the average over the unique pairs, where the range for i is 2–200, and j ⬍ i, or the lower third of the pair-wise symmetric matrix. We thank Dr. R. Purcell for providing the HBV inocula, Dr. J. S. Finlayson for critical review of the manuscript, and Dr. M. Farshid for helpful discussion. Zhang et al.

24. Galun, E., Eren, R., Safadi, R., Ashour, Y., Terrault, N., Keeffe, E. B., Matot, E., Mizrachi, S., Terkieltaub, D., Zohar, M., et al. (2002) Hepatology 35, 673–679. 25. Heijtink, R., Paulij, W., van Bergen, P., van Roosmalen, M., Rohm, D., Eichentopf, B. & Muchmore, E. (1999) J. Gen. Virol. 80, 1529–1535. 26. Klinkert, M. Q., Theilmann, L., Pfaff, E. & Schaller, H. (1986) J. Virol. 58, 522–525. 27. Neurath, A. R., Seto, B. & Strick, N. (1989) Vaccine 7, 234–236. 28. Ogata, N., Ostberg, L., Ehrlich, P. H., Wong, D. C., Miller, R. H. & Purcell, R. H. (1993) Proc. Natl. Acad. Sci. USA 90, 3014–3018. 29. Glebe, D., Urban, S., Knoop, E. V., Cag, N., Krass, P., Grun, S., Bulavaite, A., Sasnauskas, K. & Gerlich, W. H. (2005) Gastroenterology 129, 234–245. 30. Legler, K., Strohmeyer, H., Ritter, S., Gerlich, W. H. & Thomssen, R. (1983) Dev. Biol. Stand. 54, 179–189. 31. Purcell, R. H. & Gerin, J. L. (1975) Am. J. Med. Sci. 270, 395–399. 32. Milich, D. R., Hughes, J. L., McLachlan, A., Langley, K. E., Thornton, G. B. & Jones, J. E. (1990) J. Immunol. 144, 3535–3543. 33. Milich, D. R., Jones, J. E., McLachlan, A., Bitter, G., Moriarty, A. & Hughes, J. L. (1990) J. Immunol. 144, 3544–3551. 34. Peterson, D. L., Paul, D. A., Lam, J., Tribby, I. I. & Achord, D. T. (1984) J. Immunol. 132, 920–927. 35. Neurath, A. R., Kent, S. B. & Strick, N. (1986) J. Gen. Virol. 67, 453–461. 36. Pontisso, P., Ruvoletto, M. G., Gerlich, W. H., Heermann, K. H., Bardini, R. & Alberti, A. (1989) Virology 173, 522–530. 37. Young, M. D., Rosenthal, M. H., Dickson, B., Du, W. & Maddrey, W. C. (2001) Vaccine 19, 3437–3443. 38. Gripon, P., Rumin, S., Urban, S., Le Seyec, J., Glaise, D., Cannie, I., Guyomard, C., Lucas, J., Trepo, C. & Guguen-Guillouzo, C. (2002) Proc. Natl. Acad. Sci. USA 99, 15655–15660. 39. Shih, J. W., Cote, P. J., Jr., Dapolito, G. M. & Gerin, J. L. (1980) J. Virol. Methods 1, 257–273. 40. Lozier, J. N., Tayebi, N. & Zhang, P. (2005) Blood 105, 1029–1035. 41. Berman, H. M., Westbrook, J., Feng, Z., Gilliland, G., Bhat, T. N., Weissig, H., Shindyalov, I. N. & Bourne, P. E. (2000) Nucleic Acids Res. 28, 235–242. 42. Frishman, D. & Argos, P. (1995) Proteins Struct. Funct. Genet. 23, 566 –579. 43. Dixon, A. M., Venable, R. M., Pastor, R. W. & Bull, T. E. (2002) Biopolymers 65, 284–298. 44. Brooks, B. R., Bruccoleri, R. E., Olafson, B. D., States, D. J., Swaminathan, S. & Karplus, M. (1983) J. Comp. Chem. 4, 187–217.

MEDICAL SCIENCES

1. Zuckerman, J. N. & Zuckerman, A. J. (2000) J. Infect. 41, 130–136. 2. Tiollais, P. & Buendia, M. A. (1991) Sci. Am. 264 (4), 116–123. 3. Heermann, N., Goldmann, U., Schwartz, W., Seyffarth, T., Baumgarten, H. & Gerlich, W. (1984) J. Virol. 52, 396–402. 4. Courouce´, A. M., Lee, H., Drouet, J., Canavaggio, M. & Soulier, J. (1982) Dev. Biol. Stand. 54, 527–534. 5. Arauz-Ruiz, P., Norder, H., Robertson, B. H. & Magnius, L. O. (2002) J. Gen. Virol. 83, 2059–2073. 6. Norder, H., Courouce´, A. M. & Magnius, L. O. (1994) Virology 198, 489–503. 7. Okamoto, H., Tsuda, F., Sakugawa, H., Sastrosoewignjo, R. I., Imai, M., Miyakawa, Y. & Mayumi, M. (1988) J. Gen. Virol. 69, 2575–2583. 8. Stuyver, L., De Gendt, S., Van Geyt, C., Zoulim, F., Fried, M., Schinazi, R. F. & Rossau, R. (2000) J. Gen. Virol. 81, 67–74. 9. Norder, H., Courouce´, A. M., Coursaget, P., Echevarria, J. M., Lee, S. D., Mushahwar, I. K., Robertson, B. H., Locarnini, S. & Magnius, L. O. (2004) Intervirology 47, 289–309. 10. Miyakawa, Y. & Mizokami, M. (2003) Intervirology 46, 329–338. 11. Cui, C., Shi, J., Hui, L., Xi, H., Zhuoma, Quni, Tsedan & Hu, G. (2002) J. Gen. Virol. 83, 2773–2777. 12. Hannoun, C., Norder, H. & Lindh, M. (2000) J. Gen. Virol. 81, 2267–2272. 13. Kurbanov, F., Tanaka, Y., Fujiwara, K., Sugauchi, F., Mbanya, D., Zekeng, L., Ndembi, N., Ngansop, C., Kaptue, L., Miura, T., et al. (2005) J. Gen. Virol. 86, 2047–2056. 14. Morozov, V., Pisareva, M. & Groudinin, M. (2000) Gene 260, 55–65. 15. Schaefer, S. (2005) J. Viral Hepatitis 12, 111–124. 16. Cooreman, M. P., Leroux-Roels, G. & Paulij, W. P. (2001) J. Biomed. Sci. 8, 237–247. 17. Ghany, M. G., Ayola, B., Villamil, F. G., Gish, R. G., Rojter, S., Vierling, J. M. & Lok, A. S. (1998) Hepatology 27, 213–222. 18. Protzer-Knolle, U., Naumann, U., Bartenschlager, R., Berg, T., Hopf, U., Meyer zum Buschenfelde, K. H., Neuhaus, P. & Gerken, G. (1998) Hepatology 27, 254–263. 19. Gripon, P., Cannie, I. & Urban, S. (2005) J. Virol. 79, 1613–1622. 20. Ko ¨ck, J., Nassal, M., MacNelly, S., Baumert, T. F., Blum, H. E. & von Weizsa¨cker, F. (2001) J. Virol. 75, 5084–5089. 21. Bhatnager, P. K., Papas, E., Blum, H. E., Milich, D. R., Nitecki, D., Kareb, M. J. & Vyas, G. N. (1982) Proc. Natl. Acad. Sci. USA 79, 4723–4727. 22. Dreesman, G. R., Sanchez, Y., Ionescu-Matiu, I., Sparrow, J. T., Six, H. R., Peterson, D. L., Hollinger, F. B. & Melnick, J. L. (1982) Nature 295, 158–160. 23. Eren, R., Ilan, E., Nussbaum, O., Lubin, I., Terkieltaub, D., Arazi, Y., Ben-Moshe, O., Kitchinzky, A., Berr, S., Gopher, J., et al. (2000) Hepatology 32, 588–596.

Zhang et al.

PNAS 兩 June 13, 2006 兩 vol. 103 兩 no. 24 兩 9219