The hepatitis B virus (HBV) surface proteins

Transcription

1 Impact of Hepatitis B Virus Surface Protein Mutations on the Diagnosis of Occult Hepatitis B Virus Infection Mira El Chaar, 1 Daniel Candotti, 2 R. Anthony Crowther, 3 and Jean Pierre Allain 1 Genotype D occult hepatitis B virus (HBV) infections (OBIs) have a high frequency of amino acid substitutions in the major hydrophilic region of the small surface protein (S protein). This possibly reflects an escape mutation mechanism to evade detection by the host immune system. Mutations may also impact the detection of hepatitis B surface antigen (HBsAg) by commercial assays. To test these hypotheses, 20 recombinant HBV genotype D surface proteins from OBI carriers with or without antibody to hepatitis B surface antigen (anti-hbs) were expressed in yeast. Recombinant surface protein (rs protein) variants were nonreactive with autologous anti-hbs but reacted weakly with vaccine-induced anti-hbs supporting an immune escape mechanism. rs protein variants tested with a wide range of HBs antibodies, and HBsAg commercial assays showed significantly lower antigenic reactivity in anti-hbs carriers than in donors with antibody to hepatitis B core antigen (anti-hbc) only. Eight out of 10 recombinant variants from anti-hbs carriers reacted weakly or were nonreactive with antibodies to HBs as well as with qualitative and quantitative commercial HBsAg assays, whereas eight out of 10 anti-hbc only plasmas were fully reactive. rs proteins with substitutions of wild-type cysteine at positions 121, 124, and 137 were nonreactive or showed poor reactivity. However, mutation of cysteine 147 did not alter reactivity compared with controls. Restoration of cysteines 124 and 137 by sitedirected mutagenesis improved antigenic reactivity. Conclusion: Escape mutation is a mechanism associated with OBI, which also leads to decreased reactivity in HBsAg detection assays. Performance of commercial assays would be improved by the incorporation of OBI mutants in reagent development. (HEPATOLOGY 2010;52: ) The hepatitis B virus (HBV) surface proteins consist of three structurally related large, middle, and small envelope proteins. The epitopes Abbreviations: anti-hbc, antibody to hepatitis B core antigen; anti-hbs, antibody to hepatitis B surface antigen; EIA, enzyme immunoassay; HBsAg, hepatitis B surface antigen; HBV, hepatitis B virus; MHR, major hydrophilic region; OBI, occult hepatitis B infection; PBS, phosphate-buffered saline; PBST, phosphate-buffered saline/tween-20; PCR, polymerase chain reaction; rs protein, recombinant surface protein; S protein, small surface protein; S/CO, sample/cutoff ratio; SDM, site-directed mutagenesis. From the 1 Division of Transfusion Medicine, Department of Haematology, University of Cambridge, Cambridge, UK; the 2 National Health Service Blood and Transplant, Cambridge Blood Centre, Cambridge, UK; and the 3 Medical Research Council, Laboratory of Molecular Biology, Cambridge, UK. Received May 9, 2010; accepted July 21, Supported in part by Bio-Rad (Marnes-la-Coquette, France) and by the National Health Service Blood and Transplant (Cambridge, UK). M. E. C. was supported in part by a Ph.D. grant from the National Center for Scientific Research (Beirut, Lebanon). Address reprint requests to: Professor Jean Pierre Allain, Department of Haematology, Cambridge Blood Centre, Long Road, Cambridge, CB20PT, UK. jpa1000@cam.ac.uk; Fax: (44) Copyright VC 2010 by the American Association for the Study of Liver Diseases. View this article online at wileyonlinelibrary.com. DOI /hep Potential conflict of interest: Nothing to report. of small surface protein (S protein) are formed by the complex folding of the major hydrophilic region (MHR) involving cysteine residues at key positions resulting in the formation of loop-like structures that constitute the a determinant. S protein monomers aggregate in the endoplasmic reticulum and are secreted into plasma from the host cell as pseudoparticles. 1 The implementation of nucleic acid testing for HBV DNA has uncovered infected, apparently healthy blood donors escaping hepatitis B surface antigen (HBsAg) detection: occult HBV infection/carriage (OBI). OBI is characterized by a lack of reactivity with the most sensitive HBsAg assays and the presence of very low levels of HBV DNA in circulation. 2 The genesis of OBI is largely unknown, but low-level viral replication in most OBI carriers 3,4 might be related to imperfect immune control of HBV replication, as suggested by the high frequency of amino acid substitutions observed in surface proteins of genotype D strains and the presence of antibody to hepatitis B surface antigen (anti-hbs) in approximately 50% of OBI 1600

2 HEPATOLOGY, Vol. 52, No. 5, 2010 EL CHAAR ET AL Table 1. Primers Used for Cloning and SDM Experiments Primers Sequence (5 0 to 3 0 ) Amino Acid Targeted Position Cloning Primers PIC3.5K-F ACTCTACG TAACCATGGARAACAYMACATCAGGA PIC3.5K-R CCAGGGAATTCTCATTTTTCGAACTGCGGGTGGCTCCACTTGTCGTCATCGTCCTTGTAGTCATGGTG ATGGTGATGATGGGCCGCAATGTATACCCAVAGACABAAGAA SDM primers BR2-124-F ATGCAGAACCTGCACGCACGACTACTGCTCAAG 124 BR2-124-R CTTGAGCAGTAGTCGTGCAGGTTCTGCAT BR2-137-F ATCCCTCCTGTTGCTGTACCAAACCTTCG 137 BR2-137-R CGAAGGTTTGGTACAGCAACAGGAGGGAT BR2-124/137-F ACCATGCAGAACCTGCACGACTACTGCTCAAGGAACCTCTATGTATCCCTCCTGTTGCTGTACCAAA BR2-124/137-R TTTGGTACAGCAACAGGAGGGATACATAGAGGTTCCTTGAGCAGTAGTCGTGCAGGTTCTGCATGGT V5-147-F TTCGGACGCAAATTGCACCTGTATTCCCATC 147 V5-147-R GATGGGAATACAGGTGCAATTTGAGTCCGAA R1-145-F AAACTTTTGGACGGAAATTGCACCTGTATT 145 R1-145-R AATACAGGTGCAATTTCCGTCCAAAAGTTT V1-F AACCTGCATGACTACTGCTCAAGGAACCTCTATGTATCCCTCCTGTTGC 133,134 V1-R GCAACAGGAGGGATACATAGAGGTTCCTTGAGCAGTAGTCATGCAGGTT donors. 3 A variety of mutations has been identified in the HBV S proteins that potentially affects in vitro antigen detection, immune response recognition, HBV infectivity, cell tropism, and virion morphogenesis. 5-7 The aim of this study was to assess the effect of mutations observed in OBI with and without anti- HBs on S protein immune reactivity. For that purpose, recombinant surface protein (rs protein) of occult HBV strains were expressed in yeast. Their reactivity with autologous (patient s own) and heterologous antibodies as well as with a panel of commercial HBsAg screening assays used for screening blood donors or patients was studied. Patients and Methods Patient Samples. Plasma samples were obtained from 20 blood donors with confirmed HBV genotype D infection from Poland (Warsaw, [n ¼ 6]), Spain (Madrid, Valladolid, Valencia, [n ¼ 6]), Italy (Turin, Rome, [n ¼ 5]), Switzerland (Bern, [n ¼ 2]), and South Africa (Cape Town, [n ¼ 1]). 8 These samples were selected from a total of 56 strains (43 in Candotti et al. 8 [six from Switzerland, five from South Africa, 25 carrying anti-hbs]) presenting at least one MHR substitution, on the basis of the presence of amino acid substitutions in the MHR. The selected strains (10 antibody to hepatitis B core antigen [anti- HBc] only, 10 anti-hbs positive) were representative for the number of substitutions (4.8 and 5.1 for anti- HBc strains and 7.3 and 7.4 for anti-hbs overall and selected, respectively). The overall difference between anti-hbs positive or anti-hbs negative groups was significant (P ¼ 0.037). However, strains with mutated cysteines were preferentially chosen in view of their particular importance for MHR conformation. Two wild-type genotype D strains from individuals with chronic HBV infection were used as controls. Construction of Expression Vector and Transformation Into Pichia pastoris. The S gene of the control and variant samples was amplified. The sense primer (PIC3.5K-F) was designed to include SnaB1 and the ATG start codon in the context of a Kosak sequence, and the antisense primer (PIC3.5K-R) included the EcoRI restriction site, a histidine tag, flag tag, and strep tag II sequences in addition to a stop codon. In some experiments, a V5 tag was also included (Table 1). The amplified product was obtained following a second round of amplification containing 1 Expand High Fidelity buffer (2.5 mm MgCl 2, 0.2 mm dntps, 0.4 lm each primer) and 2.5 units of Expand High Fidelity Enzyme (Roche, Mannheim, Germany). The polymerase chain reaction (PCR) product of the S gene was digested and ligated into SnaBI-EcoRI treated yeast expression vector ppic3.5k (Invitrogen, Paisley, UK). The recombinant vector was linearized by digestion with SacI and transformed in Pichia pastoris Km71 strain by way of electroporation according to the manufacturer s instructions (Invitrogen). P. pastoris was selected because it provided high yield and was used to produce HBV vaccine. In addition, six strains were expressed in both yeast and mammalian cell systems to compare reactivity. Site-Directed Mutagenesis by Way of Overlap Extension PCR. Site-directed mutagenesis (SDM) was performed using a two-step PCR procedure to correct mutations at critical positions that may have caused alterations in the HBsAg conformation. Two reactions were performed with a primer pair (PIC3.5K-F/SDM-

3 1602 EL CHAAR ET AL. HEPATOLOGY, November 2010 R, PIC3.5K-R/SDM-F) using 50 ng of the original extracted mutated DNA. To obtain a full-length mutated fragment, an equimolar concentration (100 ng) of the two PCR products was used as a template in the second amplification with PIC3.5K forward and reverse primers under the conditions described above (Table 1). The PCR product was cloned in PIC3.5 vector for expression. Expression and Purification of rs Proteins. Mut S phenotype integrated into the grown colonies with the AOX1 locus of the yeast genome were selected and grown in 1 L of buffered glycerol complex medium to an optical density over 6 at 600 nm. Cells were induced with 1% methanol by replacing the buffered glycerol complex medium with buffered methanol complex medium. Methanol was added at a concentration of 1% every 24 hours during the induction phase (up to 72 hours). Cultures were centrifuged, and the pelleted cells were resuspended in a 25-mM Tris buffer (ph 7.6) and subjected to high pressure (30K psi) in a cell disrupter system (Constant System Ltd., Northants, UK). After centrifugation at 17,500g for 30 minutes at 4 C, supernatants were clarified using a 45- lm filter and subjected to ultrafiltration through 300,000 nominal molecular weight cutoff hollow fibers (GE Healthcare, Slough, UK). The rs proteins were further purified using an anti-flag M2 affinity gel (Sigma, Gillingham, UK) and eluted with 100 lg/ml of flag peptide. The eluted proteins were analyzed on 12% sodium dodecyl sulfate polyacrylamide gel electrophoresis and stained with Coomassie blue to estimate purity. Detection of the recombinant proteins was performed by way of western blot analysis using monoclonal anti-flag M2 antibody (Sigma) and confirmed with anti-his antibody (GenScript, Piscataway, NJ) and StrepMAB-classic antibody (IBA, Gottingen, Germany). Quantification of rs Proteins. Densitometry was performed to calculate the concentration of rs protein by comparison with known bovine serum albumin standards using ImageJ software ( gov/ij/). Total protein concentration was measured by way of bicinchoninic protein assay (Pierce, Rockford, IL), and purity of the HBsAg protein was calculated by dividing the rs protein concentration by the total concentration of protein. An additional method was used to confirm the concentration of the various recombinant HBsAg based on StrepTactin-coated plates (IBA). Immunolabeling and Negative Staining Transmission Electron Microscopy. To visualize whether the rs proteins were being produced in the form of pseudoparticles, immunolabeling with polyclonal antibody to HBsAg (Abcam, Cambridge, UK) or monoclonal anti- V5 antibody (Invitrogen) was performed. Immunoelectron microscopy was performed as follows; 2 ll of 100 lg/ml of rs protein suspended in Tris-buffered saline (ph 7.4) were applied for 1 minute to a glowdischarged coated grid. The grid was blocked with 0.1% gelatine/phosphate-buffered saline (PBS) for 10 minutes and treated with primary antibody at a concentration of 1:100 for 60 minutes. The grid was washed with 0.1% gelatin/pbs followed by 1-hour incubation with 10 nm gold-conjugated anti-rabbit (Sigma) or anti-mouse antibody (BB International, Agar Scientific, Stansted, UK) (1/20 dilution). The grid was stained with 1% uranyl acetate solution for 10 seconds. The grids were imaged at 40,000 nominal magnification on EM 208 S electron microscope (FEI, Eindhoven, The Netherlands). Capture Enzyme-Linked Immunosorbent Assay of rs Proteins. StrepTactin-coated plates (IBA) were coated with 1 lg/ml of rs protein diluted with Trisbuffered saline buffer for 1 hour at room temperature. Plates were washed five times with 0.1% PBS/Tween 20 (PBST) at room temperature and blocked with 4% bovine serum albumin/0.1% PBST for 2 hours. Plasma from donor samples diluted 1:100 with 2% bovine serum albumin/0.1% PBST was added to the plates and allowed to incubate for 1 hour at room temperature. Tests were performed in duplicate. After washing with 0.5% PBST, 100 ll of 1:50,000 diluted horseradish peroxidase conjugated anti-human Fc immunoglobulin G monoclonal antibody (Sigma) was added per well and incubated at room temperature for 1 hour. Plates were washed with 0.5% PBST, and 100 ll 3,39,5,59-tetramethylbenzidine substrate (Pierce, Cramlington, UK) was added per well, followed by incubation at room temperature in darkness for 20 minutes. The reaction was stopped by the addition of 100 ll of 1 M sulfuric acid. Average optical densities at 450 versus 630 nm were used to calculate sample/ cutoff ratio (S/CO, index value). The cutoff was determined as the mean absorbance of 6 HBV/anti-HBs negative plasma samples plus 6 standard deviations. Samples with index values above 1 were considered reactive. Reactivity of the rs proteins was tested with six purified monoclonal antibodies and one polyclonal antibody. The monoclonal antibodies targeted either linear epitopes (Ab 2, Ab3, and P2D3) or conformational epitopes (Ab1, D2H5, and H3F5). 9 Polysorp 96-microwell plates (Fisher, Loughborough, UK) were coated with the antibodies at 2.5 lg/ml overnight at

4 HEPATOLOGY, Vol. 52, No. 5, 2010 EL CHAAR ET AL C. Plates were washed and blocked as above. The rs proteins diluted in Tris-buffered saline (1 lg/ml) were added and incubated at room temperature for 2 hours following washing with 0.1% PBST; plates were incubated with monoclonal anti-flag M2 antibody (1: 5,000) horseradish peroxidase labeled for 2 hours. Flag peptide was used as a negative control. rs Protein Testing with HBsAg Commercial Assays. rs proteins were tested for the detection of HBsAg with 10 different commercial assays, including two quantitative chemoluminescent immunoassays (Architect HBsAg [Abbott, Delkenheim, Germany] and Elecsys HBsAg II [Roche]) and eight qualitative assays (Monolisa HBsAg Ultra [Bio-Rad, Marnes-la- Coquette, France]; Hepanostika HBsAg Ultra [bio- Merieux, Marcy-I Etoile, France]; Murex HBsAg version 3 [Abbott]; Bioelisa HBsAg Colour [Biokit, Barcelona, Spain]; SD HBsAg 3.0 [Standard Diagnostics, Suwon, Korea]; Determine HBsAg [Abbott]; DRW-HBsAg [DRW, Sunnyvale, CA]; and HBsAg Fast [Standard Diagnostics]). rs proteins were diluted to 1 lg/ml in Tris-buffered saline and tested according to the manufacturers instructions. Statistical Analysis. Statistical analysis with an unpaired t test was performed using GraphPad Prism version 4.00 (GraphPad Software, San Diego, CA). A two-tailed P value of 0.05 was considered significant. Results OBI Sample Characterization. Among the 20 selected OBI cases, 19 (95%) donors were men whose ages ranged between 32 and 73 years, with a median of 55 years (Table 2). Donors were clinically asymptomatic and had normal alanine aminotransferase levels (<50 IU/L), except donor BR1, who had an alanine aminotransferase level of 85 IU/L. None of the donors was vaccinated against HBV, but 10 of them had low plasma levels of anti-hbs ranging between 11 and 204 IU/L (median, 13 IU/L). All samples were anti-hbc positive and carried a viral load below 70 IU/mL. When compared with a genotype D consensus amino acid S sequence derived from 25 HBsAg-positive patients (sequences available in GenBank, see Candotti et al. 8 ), the 20 OBI strains investigated had 1-27 amino acid substitutions (mean, 12; median, 10) in the MHR, whereas fewer substitutions were seen in the neighboring regions (mean, 3; median, 3). The mean number of substitutions was higher in anti- HBs positive samples (7.4) than in anti-hbc only samples (5.1). Substitutions of the wild-type cysteines Table 2. Donor Information and Hepatitis B Markers Samples Sex Age Viral Load (IU/mL) Anti-HBc Anti-HBs (IU/L) Alanine Aminotransferase Level (IU/L) P10 M Pos Neg 27 P15 M Pos Neg 20 P17 M 42 <1 Pos Neg 39 T6 M 56 2 Pos Neg 23 T3 M Pos Neg 14 T8 M Pos Neg 17 T5 M Pos Neg 15 VA 1 M 44 <1 Pos Neg 41.5 V11 M Pos Neg ND CT2 M Pos Neg ND V1 M Pos V5 M 47 <1 Pos V6 F Pos P1 M Pos Pos 32 P2 M 50 <1 Pos P5 M 42 <1 Pos MD1 M Pos R1 M 51 <1 Pos BR1 M Pos BR2 M Pos Abbreviations: F, female; M, male; ND, not determined; Neg, negative; Pos, positive. at positions 121, 124, and 137 were identified in anti- HBs positive donors (MD1, BR2, and R1). In addition to substitution of cysteine at position 137, sample R1 had the wild-type glycine at position 145 replaced by arginine as well as 10 other substitutions along the MHR. Cysteine 147 substitution was seen in both anti-hbs negative (VA1, V11) and anti-hbs positive plasmas (V5 and MD1). The two control rs proteins prepared according to the same protocol as the variants were not identical (Table 3), although reactivity was similar but generally slightly higher in the mutated control (Tables 4 and 5). Characterization of rs Proteins Produced in P. pastoris. rs proteins were produced in yeast cells and secreted at a very low level. The product of the S gene appeared on 12% sodium dodecyl sulfate polyacrylamide gel electrophoresis as one major band of approximately 48 kda, which corresponded to rs protein dimers and was confirmed on western blot analysis using anti-flag M2 and anti-his antibodies (Fig. 1). The rs proteins were confirmed to be nonglycosylated through treatment with endonuclease H (data not shown). The purity of the eluted preparation ranged between 70% and 90% (median, 85%). Electron microscopic analysis of purified rs proteins was performed to determine whether pseudoparticles were formed. rs proteins were released as apparently spherical particles of two main sizes (20-30 nm and nm) (Fig. 2A). Both types of pseudoparticles were

5 Table 3. Amino Acid Mutations of OBI Variant in the MHR Amino Acid Sequence of MHR Sample City of origin Subtype Anti-HBs Consensus ayw Y O G M L P V C P L I P G S S T T S T G P C R T C T T P A O G T S M Y P S C C C T K P S D G N C T C I P Control 1 ayw Neg M N T Control 2 ayw Neg P10 Warsaw, Poland adw Neg R K V K N Y L P15 Warsaw, Poland ayw Neg K L N R P P17 Warsaw, Poland ayw Neg Q N T R I M A T6 Turin, Italy adw Neg K S E T3 Turin, Italy ayw Neg H S P T T8 Turin, Italy ayw Neg P E T5 Turin, Italy ayw Neg A VA 1 Valladolid, Spain ayw Neg I A Y V11 Valencia, Spain ayw Neg N R E N R E Y CT2 Cape Town, South Africa ayw Neg R N T V P K I H Y V1 Valencia, Spain adw Pos S H V N K L H V5 Valencia, Spain ayw Pos I A I R V6 Valencia, Spain ayr Pos A P1 Warsaw, Poland ayw Pos A A E P2 Warsaw, Poland ayw Pos T A M K F V P5 Warsaw, Poland ayw Pos R G I R F L MD1 Madrid, Spain ayw Pos Q N R S S P M N R D P I T F I E Y R1 Rome, Italy adw Pos N A T K V S H W L L R L BR1 Bem, Switzerland ayw Pos R P P N K K S S I Q N T F Y BR2 Bem, Switzerland ayw Pos A Y Y Amino acid substitutions are indicated with one-letter codes. Substitutions at position 122 are indicative of serotype. Abbreviations: Neg, negative; Pos, positive.

6 HEPATOLOGY, Vol. 52, No. 5, 2010 EL CHAAR ET AL Table 4. Reactivity of rs Proteins with Various Anti-HBs Expressed as S/CO rs Protein Anti-HBs rs Protein/Donor Autologous Plasma rs Protein/Vaccines Polyclonal Anti-HBs- Immunoglobulin G Sample/Monoclonal Ab to Linear Epitopes Sample/Mono-Polyclonal Ab to Conformational Epitopes Ab-2 Ab-3 P2D3 Ab-1 D2H5 H3F5 Ab-4 Control 1 Neg NA Control 2 Neg NA P10 Neg P15 Neg P17 Neg NA T6 Neg NA T3 Neg T8 Neg T5 Neg NA VA1 Neg V11 Neg NA CT2 Neg V1 Pos NA V5 Pos V6 Pos NA P1 Pos P2 Pos P5 Pos MD1 Pos R1 Pos NA BR1 Pos BR2 Pos P value NA Abbreviations: Ab, antibody; NA, not available; Neg, negative; Pos, positive. confirmed to correspond to HBV surface proteins by way of immunoelectron microscopy using anti-tag and polyclonal antibody to HBsAg (Fig. 2C,D). Measurement of the particle size using ImageJ software revealed a wide variation in particle diameter (range, nm): 52% of particles had a diameter of nm, 22% had a diameter of nm; 24% had a diameter <19 nm, and 8% had a diameter of >60 nm. In addition, in samples BR2 and V1, some apparently incomplete particles with a crescent or V-like morphology were seen, but not in the controls or in other rs protein variants (Fig. 2B). Reactivity of S Protein Variants with Autologous/ Heterologous Plasmas. The panel of purified rs proteins was tested for reactivity against human antibodies to HBsAg. Plasmas from OBI donors with cocirculating HBV DNA and anti-hbs were tested against the autologous rs proteins produced from their own circulating strains. None of the donor plasmas reacted with autologous variants (Table 4). However, anti-hbc only plasma from donor P15 weakly reacted with its own strain. In contrast, when rs variants were tested with anti-hbs to genotype A2 obtained from an HBV-vaccinated individual, strains cocirculating with anti-hbs reacted poorly in comparison with rs proteins from anti-hbc only plasmas (P ¼ ) with the exception of the variant from donor P17. Reactivity Patterns of Various Anti-HBs Against OBI rs Variant Proteins. Probing of variant rs protein reactivity with a panel of linear or conformational epitope-specific antibodies revealed several interesting features (Table 4). First, irrespective of the type or localization of epitopes, 10 recombinant variant proteins were either nonreactive or weakly reactive (mean S/CO <10), eight of which were from anti-hbs positive donors (V1, V5, P2, P5, MD1, R1, BR1, BR2) and two of which were from anti- HBc only OBI donors (P17, CT2). Of the weakly reactive proteins, the lowest readouts were from strains with substituted cysteine at position 137 (R1, C137W and G145R; BR2, C137Y and C124Y), those with two mutated cysteines (MD1, C121S and C147Y; BR2, C124Y and C137Y), or those with multiple mutations in antigenic loops 1 and 3, including CT2 (10 substitutions) and BR1 (15 substitutions). Variant protein V1, with two doublet substitutions at positions and , was one of the least reactive. In contrast, substitution of cysteine 147 did not appear to affect reactivity, either alone or in association with few other amino acid substitutions. The cumulative reactivity of surface antigen antibodies was significantly lower in anti- HBs positive plasma (mean S/CO, 6.5) than anti- HBc samples (mean S/CO, 13.0) (P < ).

7 1606 EL CHAAR ET AL. HEPATOLOGY, November 2010 Table 5. Detection of rs Proteins by Qualitative and Quantitative HBsAg Commercial Assays EIA (Qualitative Assays)* CIA (Quantitative Assays)y Rapid Test (Qualitative Assays)z rs Proteins Anti-HBs Control 1 Neg þþ þþ þþ Control 2 Neg þþ þþ þþ P10 Neg þþ þ þ P15 Neg þ þ þ P17 Neg <0.05 <0.05 þ T6 Neg þþ þþ þþ T3 Neg þþ þþ T8 Neg þþ þþ þþ T5 Neg þþ þþ þþ VA1 Neg þ þ þ V11 Neg CT2 Neg V1 Pos <0.05 <0.05 V5 Pos V6 Pos þþ þþ þþ P1 Pos þþ þ þþ P2 Pos <0.05 <0.05 P5 Pos þþ þ MD1 Pos <0.05 <0.05 R1 Pos <0.05 <0.05 BR1 Pos <0.05 <0.05 BR2 Pos <0.05 <0.05 P value Abbreviations: CIA, chemiluminescent immunoassays; EIA, enzyme immunoassays; Neg, negative; Pos, positive. *Reactivity is measured as S/CO. HBsAg concentration is measured in IU/mL. Rapid test reactivity was determined visually: strong signal (þþ), weak signal (þ), no signal ( ). P value comparing S/COs of anti-hbs^positive and anti-hbs^only plasmas. The number of reactive samples was used to calculate the P value of the rapid tests. Immunoreactivity of rs Protein Variants with Commercial HBsAg Assays. Ten HBsAg commercial assays (five qualitative enzyme immunoassays [EIAs], two quantitative chemoluminescent immunoassays, and three rapid tests) were used to test assay reactivity against the panel of 20 recombinant proteins (Table 5). These commercial assays used at least two antibodies: one for HBsAg capture, the other for detection. Results were essentially consistent with those obtained with the panel of individual monoclonal antibodies. rs protein variants originating from anti-hbs positive OBI samples reacted at a significantly lower level than those from anti-hbc only OBI donors (P < 0.001) (Table 5). However, two rs proteins (V6 and P1) from the anti-hbs group were highly reactive with all EIAs, whereas the remaining eight proteins had lower levels of reactivity or were nonreactive with most immunoassays. Among the 10 variants originating from donors with undetectable anti-hbs, two reacted poorly with all tests (P17 and CT2) in comparison with the other eight proteins, which were highly reactive with most assays. The variant proteins V11 and T3 were well recognized by three and two EIAs, respectively, but not by the other assays. Remarkably, both quantitative assays performed poorly, positively reacting with seven of 20 variant S proteins (35%), six of which were among the 10 strains from anti-hbc positive OBI donors (P ¼ 0.04). The three HBsAg rapid tests essentially detected the same seven variants, although three additional variants were detected by either one or two rapid tests. The total reactivity of all rapid tests with variants from anti-hbs positive donors was significantly lower than those from anti-hbc only donors (P ¼ 0.005). Among six proteins expressed in the mammalian system, only BR2 showed different reactivity from the one expressed in the yeast. Reactivity of Six Variant rs Proteins Partially Corrected by SDM. In order to determine the specific role played by individual amino acid substitutions in the antigenic reactivity of the MHR, mutated cysteines and selected additional amino acids were restored to the wild-type residue by SDM. In sample BR2, mutated cysteines 124 and 137 were restored to wild-type, both individually and collectively. With both the monoclonal antibodies and the commercial EIAs, reactivity was recovered to some degree (S/CO<5) with single cysteine restoration (Fig. 3), and the highest reactivity was observed when both cysteines were restored. Despite

8 HEPATOLOGY, Vol. 52, No. 5, 2010 EL CHAAR ET AL Fig. 1. Sodium dodecyl sulfate polyacrylamide gel electrophoresis of purified rs proteins. (A) Coomassie blue staining of the control surface protein under reduced conditions following purification with anti- FLAG M2 affinity gel. (B) Western blot of the control protein using a monoclonal anti-flag M2 antibody at different dilutions. rs proteins form a dimer of approximately 48 kda. improving the mean S/CO of the monoclonal antibodies from 1.7 for the mutant to 7.6 following restoration of the two mutated cysteines, this was still a lower S/ CO than that observed in controls (15.8). The mean reactivity for the five EIAs examined was 1.3 with the dual mutation and reached 7.1 following correction, but the control S proteins had a mean S/CO of Restoration of cysteine at position 147 improved the reactivity of recombinant V5 with EIAs from a mean S/CO of 2.0 to 5.5. However, this effect was not seen in recombinants VA1, V11, or MD1 (data not shown). Restoration of glycine 145, methionine 133, and tyrosine 134 did not substantially improve reactivity with the assays, because the S/COs of R1 (11 substituted amino acids) and V1 (6 substituted amino acids) did not change following correction (data not shown). Discussion The high frequency of amino acid substitutions in the MHR of genotype D OBI strains has been described 3 and may account for escaping immune response and/or nondetection by screening assays despite the presence of detectable HBV DNA. OBIs carrying both anti-hbc and anti-hbs were considered under stronger humoral immune pressure than those carrying only anti-hbc either in recovered hosts no longer producing anti-hbs or in chronic infections having HBsAg no longer detectable, although many substitutions were found in these strains (Table 3). Direct evidence of the role of immune pressure was given by anti-hbs positive plasmas not reacting (with the exception of P15 which had a weak response) with rs proteins derived from the same plasmas. This observation directly supports the mechanism of virus escaping the host immune system (Table 4). The fact that antibodies obtained from a vaccinated individual reacted weakly with these proteins indicates that polyclonal antibodies can recognize the mutant proteins, although autologous plasmas cannot. As a corollary, it suggests that these escape mutants are antigenic and, once recognized by the immune system, specific neutralizing antibodies might be elicited, clearing them from circulation. Indirect evidence of such a mechanism is provided by follow-up samples obtained from several samples collected in Poland (P1, P2, P10, P15), where 4-24 weeks after the index samples were identified as anti-hbs positive OBI, HBV DNA was no longer detectable but anti-hbs remained, although at relatively low levels. 10 These data suggest that in individuals able to produce anti-hbs, OBI might be a temporary status until escape mutant S protein is ultimately recognized by the host immune system. Alternatively, as has been suggested, 10 DNA levels close to the assay detection limit may fluctuate so that they are detectable or not with relatively small variations. However, in some cases, HBV DNA and anti-hbs seem to persist over time, suggesting that (1) the mutated protein is produced in quantities below that required for immune recognition; or (2) despite being antigenic, the mutated protein remains hidden from immune surveillance; or (3) alternatively, some form of immune tolerance may be established between virus and host. The reactivity of recombinant proteins with specific antibodies (Table 4) or commercial assays (Table 5) indicates that some samples react at levels close to the control, whereas others react weakly or not at all, irrespective of the antibody or the assay. However, several samples were negative with one assay but weakly positive with another (S/CO >1). As a result, the first assay would be HBsAg-negative and the sample classified as OBI, whereas the testing of the same sample with an alternative assay would identify a low level of HBsAg, resulting in a diagnosis of chronic infection, unless anti-hbs is simultaneously detected. The diagnosis of OBI is therefore closely dependent on the ability of the HBsAg screening assay to recognize MHR variant epitopes. Samples identified as OBI should be retested with multiple HBsAg assays, and a diagnosis should be made only when all assays are negative. In addition, as shown in Table 5, quantitative assays not only fail to detect a large number of OBI variants but also tend to underestimate the level of HBsAg in samples with low reactivity. When analyzing the reactivity data according to the two serological groups with or without anti-hbs

9 1608 EL CHAAR ET AL. HEPATOLOGY, November 2010 Fig. 2. Electron micrograph of purified rs proteins. (A) Nonstained wild-type control 1 and (B) mutant BR2 rs protein at protein concentration of approximately 100 lg/ml. BR2 rs protein presents abnormal, incompletely formed, pseudoparticles in crescent or V shape (arrows). (C,D) Immunolabeling of rs control protein with (C) anti-hbs and (D) V5 anti-tag. Particles of large or small diameter are labeled and indicated by arrows. Scale bar, 100 nm. (Table 2), it was clear that antigenic reactivity of MHR was significantly lower in anti-hbs carriers than in anti-hbc only donors. However, in both groups, the number of substitutions was inversely correlated to antigenic reactivity, except when cysteines in specific positions were involved. The decreased or absent reactivity was found with both specific monoclonal antibodies (in some, the targeted epitope was known) (Table 4) and commercial HBsAg assays, whose antibodies or mixtures of antibodies for both capture or detection were of unknown specificity (Table 5). Not surprisingly, strains with heavily substituted MHRs (>8 amino acid differences from consensus sequence) tended to react less regardless of whether anti-hbs or only anti-hbc was present. Within each group, samples tended to have similar levels of reactivity, although some exceptions were identified. Two anti-hbc samples (P17 and CT2) reacted poorly but presented the largest number of substitutions in that group. These two strains may have been from recovered donors who were no longer producing anti-hbs. In the anti-hbs group, eight of 10 strains were weakly reactive or nonreactive; however, the two strains reacting at high levels had only a few amino acid substitutions (V6, P1). In contrast, mutations of wild-type cysteines, whether alone or accompanied by other mutations, profoundly affected epitope recognition and reactivity. Substitution of cysteines at positions 121, 124, or 137, which are known to affect the surface antigen conformation, 11 was seen only in donors with anti-hbs present (40%). Having cysteines 124 and 137 substituted in strain BR2 appeared to affect the formation of Fig. 3. Reactivity of rs protein BR2 before and after restoring mutated cysteines to wild-type by SDM. (A) Reactivity of BR2 rs protein with five HBsAg commercial EIAs (as in Table 5) is expressed as S/CO (>1.0 is considered reactive). Positive control is rs wild-type protein of control 1. (B) Reactivity of BR2 rs protein with monoclonal antibodies labeled as in Table 4, expressed as S/CO. Positive control is rs wild-type protein of control 1.

10 HEPATOLOGY, Vol. 52, No. 5, 2010 EL CHAAR ET AL pseudoparticles (Fig. 2), suggesting that external S protein conformation affected spontaneous aggregation of the S protein. This finding was further supported by a similar abnormality that occurred when this clone was expressed in mammalian cells. Variants with mutations at these positions (V5, MD1, R1, BR2) reacted poorly with all antibodies and commercial assays, with a mean S/CO value of 3.4 and 1.3, respectively. MD1, R1, and BR2 S proteins were nonreactive in the two quantitative assays. However, V5 reacted weakly (mean, 0.1 IU/mL). A reduced sensitivity for escape mutant detection was shown in the quantitative assay in comparison with the qualitative one from the same manufacturer. This may be explained either by the use of different monoclonal antibodies or by the different incubation times: the quantitative assay s 30-minute incubation time may have been too short to allow optimal presentation of epitopes to antibodies with relatively low affinity and to allow appropriate binding to the antigen. In contrast, substitution of cysteine 147 was seen in 20% of both anti-hbs and anti-hbc only donors and the mutated proteins reacted well with purified antibodies and commercial assays (mean S/ CO, 14 and 14.5, respectively). Additionally, they were reactive with quantitative assays, with a mean HBsAg concentration of 3.7 IU/mL. To demonstrate that cysteines 124 and 137 were essential for epitope presentation and assay reactivity, they were restored in the BR2 variant by SDM, both individually and as a dual rescue. The results showed that replacement of either mutant tyrosine 124 or 137 by wild-type cysteine restored some degree of immunoreactivity of BR2 rs protein (Fig. 3). However, when both mutant residues were back-mutated to wild-type cysteine, higher but still only partial reactivity was restored. It is not clear why the reactivity was not fully recovered when both cysteines were restored and whether the lack of glycosylation in the yeast system, the proper environment for protein folding, or the small number of mutations outside the a determinant had an impact on the limited reactivity of wild-type sequence-restored BR2. SDM of cysteine 147 did not increase the reactivity of the V5 variant. These results may validate the assumption that cysteine 147 in contrast to the other cysteines, which are only mutated in variants from anti-hbs positive donors might not be essential for epitope presentation of the surface protein. The results obtained confirm previous studies showing the effect of mutated cysteines on secretion and antigenicity of HBV small envelope protein. 11 The substituted G145R/A was only present in anti- HBs positive donor R1. Having been recognized 20 years ago, G145R/A is usually part of the panels used by assay manufacturers to select antibodies to capture and detect mutated HBsAg in the development of assays designed to react with mutant HBsAg. 12,13 However, R1 reacted weakly or not at all with all assays performed, with a mean S/CO of 1. The poor reactivity of recombinant R1, both as a mutant strain and after correction to wild-type R145G, strongly suggests that the reactivity was additionally affected by the presence of 12 other substitutions in the MHR, including C137Y and six hydrophilic amino acids replaced by hydrophobic residues (Table 3). Other changes could be involved in the conformational alteration of HBsAg; for example, P120S, T126I, M133I, P142L, S143L, and D144E mutations that have been reported to escape commercial assay detection 14,15 were present in CT2, V5, MD1, and BR1. These variants reacted weakly with all assays in comparison with controls. In V1, correction of M133I and Y134H by SDM did not restore the reactivity, which may be explained by the presence of another mutation acting alone or in combination with M133I to alter the conformation of the a determinant. Interestingly, the rs protein from both anti-hbc only donors (P15 and P17) and anti-hbs positive donors (V1, MD1, R1, and BR1) had a T115N substitution that created a potential new glycosylation site in addition to the wild-type glycosylation site at N146. BR2 had a third additional potential glycosylation site, at position 132. It has not yet been determined whether these neoglycosylation sites are active on rs proteins when expressed in the yeast system. A recent study showed that additional glycosylation sites might reduce antigenicity. 16 In summary, donors carrying anti-hbs tend to select multiple amino acid substitutions that escape immunologic recognition, whereas in the absence of direct antibody-selective pressure, substitutions observed in anti-hbc only OBI donors occur randomly, affecting antigenic recognition more by accumulation than selection. These findings also suggest that the previously reported classification of OBI (which indicated that HBsAg nonreactive variants were considered false OBIs) requires revision, because at the very least, infections with low viral load are clearly OBIs. 2 In many countries, nucleic acid testing is not used for screening blood units, which are tested only for the presence of HBsAg. The large number of MHR variants not described should be considered for assay development and performance evaluation. Introducing multiple antibodies that can detect a wider range of variants or an antibody that targets a linear epitope in a very conserved region not affected by

11 1610 EL CHAAR ET AL. HEPATOLOGY, November 2010 other epitopal conformational changes will enable assays to identify HBV infection otherwise diagnosed as OBI. Acknowledgment: The authors wish to thank the following collaborators who provided blood donor samples containing the HBV strains studied: Ewa Brojer, Institute of Haematology and Transfusion Medicine, Warsaw, Poland; Roberto Roig, Community Transfusion Center, Valencia, Valencia, Spain; Paola Iudicone, Company Hospital S Camillo-Forlanini, Rome, Italy; Paola Ghiazza, Blood Testing Laboratory Sant Anna Hospital, Turin, Italy; Christoph Niederhauser and Martin Stolz, Blood Transfusion Service, Bern, Switzerland; Arthur Bird, Western Province Blood Transfusion Service, Cape Town, South Africa; and Isabel González-Fraile and Lydia Bianco, Regional Blood Transfusion Center, Valladolid, Spain. We are also grateful to Laura Cox for editing the manuscript. References 1. Eble BE, Lingappa VR, Ganem D. Hepatitis B surface antigen: an unusual secreted protein initially synthesized as a transmembrane polypeptide. Mol Cell Biol 1986;6: Raimondo G, Allain JP, Brunetto MR, Buendia MA, Chen DS, Colombo M, et al. Statements from the Taormina expert meeting on occult hepatitis B virus infection. J Hepatol 2008;49: Candotti D, Allain JP. Transfusion-transmitted hepatitis B virus infection. J Hepatol 2009;51: Allain JP, Belkhiri D, Vermeulen M, Crookes R, Cable R, Amiri A, et al. Characterization of occult hepatitis B virus strains in South African blood donors. HEPATOLOGY 2009;49: Bruss V. Hepatitis B virus morphogenesis. World J Gastroenterol 2007; 13: Weber B. Genetic variability of the S gene of hepatitis B virus: clinical and diagnostic impact. J Clin Virol 2005;32: Salisse J, Sureau C. A function essential to viral entry underlies the hepatitis B virus a determinant. J Virol 2009;83: Candotti D, Gzabarczyk P, Ghiazza P, Roig R, Casamitjana N, Iudicone P, et al. Characterization of occult hepatitis B virus from blood donors carrying genotype A2 or genotype D strains. J Hepatol 2008; 49: Beale MA, Ijaz S, Tedder RS. The genetic backbone modulates the phenotype of hepatitis B surface antigen mutants. J Gen Virol 2010;91: Brojer E, Grabarczyk P, Liszewski G, Mikulska M, Allain JP, Letowska M. Characterization of HBV DNAþ/HBsAg- blood donors in Poland identified by triplex NAT. HEPATOLOGY 2006;44: Mangold CM, Streeck RE. Mutational analysis of the cysteine residues in the hepatitis B virus small envelope protein. J Virol 1993;67: Carman WF, Zanetti AR, Karayiannis P, Waters J, Manzillo G, Tanzi E, et al. Vaccine-induced escape mutant of hepatitis B virus. Lancet 1990;336: Coleman PF, Chen YC, Mushahwar IK. Immunoassay detection of hepatitis B surface antigen mutants. J Med Virol 1999;59: Van Roosmalen MH, de Jong JJ, Haenen W, Jacobs T, Couwenberg F, Ahlers-de Boer GJ, et al. A new HBsAg screening assay designed for sensitive detection of HBsAg subtypes and variants. Intervirology 2006;49: Ireland JH, O Donnell B, Basuni AA, Kean JD, Wallace LA, Lau GK, et al. Reactivity of 13 in vitro expressed hepatitis B surface antigen variants in 7 commercial diagnostic assays. HEPATOLOGY 2000;31: Wu C, Zhang X, Tian Y, Song J, Yang D, Roggendorf M, et al. Biological significance of amino acid substitutions in hepatitis B surface antigen (HBsAg) for glycosylation, secretion, antigenicity and immunogenicity of HBsAg and hepatitis B virus replication. J Gen Virol 2010;91: