Recombinant human monoclonal antibodies against ... - CiteSeerX

Journal of General Virology (2000), 81, 2451–2459. Printed in Great Britain ...................................................................................................................................................................................................................................................................................

Recombinant human monoclonal antibodies against different conformational epitopes of the E2 envelope glycoprotein of hepatitis C virus that inhibit its interaction with CD81 Tobias Allander,1, 2 Katarina Drakenberg,1 Aster Beyene,1 Domenico Rosa,3 Sergio Abrignani,3 Michael Houghton,4 Anders Widell,5 Lena Grillner2 and Mats A. A. Persson1 1, 2

Karolinska Institute, Department of Medicine1 and Department of Laboratory Medicine2, Center for Molecular Medicine (L8 : 01), Karolinska Hospital, S-171 76 Stockholm, Sweden 3 Chiron-Biocine, IRIS, Siena, Italy 4 Chiron Corporation, Department of Virology, Emeryville, CA, USA 5 Lund University, Department of Clinical Microbiology, Malmo$ University Hospital, Malmo$ , Sweden

The antibody response to the envelope proteins of hepatitis C virus (HCV) may play an important role in controlling the infection. To allow molecular analyses of protective antibodies, we isolated human monoclonal antibodies to the E2 envelope glycoprotein of HCV from a combinatorial Fab library established from bone marrow of a chronically HCV-infected patient. Anti-E2 reactive clones were selected using recombinant E2 protein. The bone marrow donor carried HCV genotype 2b, and E2 used for selection was of genotype 1a. The antibody clones were expressed as Fab fragments in E. coli, and as Fab fragments and IgG1 in CHO cells. Seven different antibody clones were characterized, and shown to have high affinity for E2, genotype 1a. Three clones also had high affinity for E2 of genotype 1b. They all bind to conformation-dependent epitopes. Five clones compete for the same or overlapping binding sites, while two bind to one or two other epitopes of E2. Four clones corresponding to the different epitopes were tested as purified IgG1 for blocking the CD81–E2 interaction in vitro ; all four were positive at 0n3–0n5 µg/ml. Thus, the present results suggest the existence of at least two conserved epitopes in E2 that mediate inhibition of the E2–CD81 interaction, of which one appeared immunodominant in this donor.

Introduction Infection with hepatitis C virus (HCV) is a global health problem, with an estimated 1–2 % of the world’s population chronically infected. Long term complications may be severe, and include liver failure and hepatocellular carcinoma. Interferon-α is the most effective pharmaceutical agent for treating the infection, with a success rate of 20–25 %. Recently, combination therapy with Ribavirin has been reported to improve the rate of complete sustained response to about 40 % (Reichard et al., 1998). Infection by HCV induces a broad cell-mediated immune response as well as a humoral response (Alter et al., 1989 ; Rehermann et al., 1996). Several reports indicate that elicited antibodies may neutralize the infectivity of the virus Author for correspondence : Mats Persson. Fax j46 8 5177 61 80. e-mail Mats.Persson!cmm.ki.se

0001-7089 # 2000 SGM

(Farci et al., 1994 ; Shimizu et al., 1994). Despite this, most infected individuals develop a chronic infection. Although antibodies elicited by natural infection are not sufficient to clear the infection or even to prevent reinfection, several reports indicate that the antibody response to HCV plays an important role in controlling the infection and modulating its course. HCV-infected patients with agammaglobulinaemia have a much more rapidly progressing disease than other patients with hepatitis C (Bjøro et al., 1994), and passively transfused antibodies to HCV have been shown to reduce the risk of HCV reinfection after liver transplantation (Feray et al., 1998). In chimpanzee experiments, repeated post-exposure treatment with polyclonal anti-HCV antibodies suppressed all signs of infection and liver disease for as long as antibodies were administered (Krawczynski et al., 1996). The lack of a reliable cell culture system for HCV has made studies of virus neutralization difficult. Two putative targets for neutralizing antibodies in the E2 envelope glycoprotein CEFB

T. Allander and others

have been suggested. The first is the hypervariable region (HVR) located in the N terminus of E2. This has been shown to encode linear epitopes that change considerably over time, resulting in new variants that escape the reactivity of previously produced, specific antibodies (Kato et al., 1993). Several reports indicate that antibodies to HVR may neutralize the virus (Farci et al., 1996 ; Shimizu et al., 1996 ; Zibert et al., 1995). During acute HCV infection, early appearance of antibodies to the HVR predicts clearance of the infection (Allander et al., 1997 ; Zibert et al., 1997). Secondly, E2 glycoprotein interacts with CD81 on human cells (Pileri et al., 1998). The capacity of antibodies to inhibit this interaction can be assessed by a competition assay, referred to as the neutralization of binding (NOB) assay. The NOB mediating site appears to be different from HVR (Rosa et al., 1996). High NOB titres were shown to correlate with protection in vaccinated chimpanzees. Furthermore, patients from a rare group that spontaneously clear chronic HCV infection have presented very high NOB titres (Ishii et al., 1998). Neutralizing antibodies of human origin could be considered for use in passive immunotherapy, in particular if broadly reactive against different strains of the virus. Preferably, such antibodies would be monoclonal in order to predict the effects in vivo ; however, a combination of several monoclonal antibodies may be superior in the case of a variable virus such as HCV. Secondly, generation of monoclonal antibodies is the most useful approach to study and characterize the immune response to the virus, provided that the antibodies isolated originate from an infected donor (Chanock et al., 1993) ; indeed, a recent report has applied this to the study of human anti-HCV antibodies (Burioni et al., 1998). In the present report, we describe the isolation of human, monoclonal anti-HCV antibodies from a combinatorial Fablibrary derived from an HCV-infected individual. The experiment was designed to result in cross-genotype reactive antibodies to the E2 envelope glycoprotein. Initially isolated as Fab clones, they were eventually expressed as IgG1 molecules in eukaryotic cells, and tested for functional anti-HCV properties.

marrow donation was found to be 2b (Widell et al., 1994) by PCR-based typing.

Library construction. Lymphocytes were isolated from the bone marrow sample using Ficoll–Paque (Pharmacia). Total RNA was extracted by the acid guanidinium thiocyanate–phenol method (Chomczynski & Sacchi, 1987). First strand cDNA synthesis utilizing oligo(dT) priming of 10 µg of RNA was performed (cDNA synthesis kit) and heavy (Fd) and light chain DNAs were PCR amplified using 5h biotinylated primers of previously published sequences for γ1 and κ chains (Kang et al., 1991 ; Persson et al., 1991). The PCR products encoding the heavy and light chains were pooled separately, phenol–chloroform extracted and ethanol precipitated. Twelve µg each of heavy and light chain DNA were gel-purified on a 2n5 % agarose gel, electroeluted, and digested with the restriction endonucleases XhoI\SpeI and SacI\XbaI, respectively (Life Technologies). The digested PCR products were subsequently gel-purified, recovered by electroelution, and ligated into the vector pComb3H after it had been digested with the corresponding enzymes and gel-purified\electroeluted as described previously for a similar vector system (Barbas et al., 1991 ; Barbas & Wagner, 1995). The resulting combinatorial library was expressed in phage, including harvesting of phage, as reported by Samuelsson et al. (1995).

Selection procedure. Four wells of a microtitre plate (Costar 3690) were coated at 4 mC overnight with 50 µl of purified recombinant E2 protein (genotype 1a), at 2n5 µg\ml, or a similarly expressed and purified E1\E2 heterodimer protein (genotype 1a) at 10 µg\ml (Spaete et al., 1992). The wells were blocked by filling completely with 5 % non-fat dry milk in PBS for 1 h at 22 mC. Phage library (50 µl ; 5i10"! c.f.u.) was then added to each well. Adsorption of phage, washing, and elution of phage was performed as described by Burton et al. (1991). Washing was done one to ten times as described below. E. coli XL-1 blue was infected with the eluted phage, an aliquot was plated, and phage was propagated after each round of panning as described by Samuelsson et al. (1995). Three separate panning experiments were performed. In the first experiment (panning series I), three rounds of panning with recombinant E2 protein were performed, and the number of washes was increased for each round (one, three and ten washes respectively). The second experiment (panning series II) consisted of a single selection with ten washes, using the same antigen. Panning series III was performed as series I, but utilizing the recombinant E1\E2 protein heterodimer.

Antigens. Recombinant E2 protein of genotype 1a and 1b, expressed in CHO cells, or a similarly expressed E1\E2 heterodimer protein (genotype 1a) were used for selection and characterization of the antibodies. The expression and purification of these recombinant proteins have been reported elsewhere (Rosa et al., 1996 ; Spaete et al., 1992).

Expression of Fab in E. coli. Fab fragments were expressed by growing ampicillin-resistant E. coli colonies containing the Fab plasmid (with the phage-derived gIII part intact, or deleted by digestion with SpeI and NheI) in SB with 50 µg\ml ampicillin and 1 % glucose, until OD '!! l 1n0 was reached. Bacteria were pelleted by centrifugation, and media exchanged to SB with 1 mM IPTG and 20 mM MgCl ; bacteria were # then resuspended and the culture incubated at room temperature on a shaker platform at 290 r.p.m. overnight. The following day, cells were spun down, the supernatant discarded, PBS added (2 % of original culture volume) and the periplasmic content released by freeze–thawing three times. Released bacterial DNA was digested with DNase I, bacterial debris pelleted by centrifugation, and the Fab-containing supernatant aliquoted to new vials. The material was kept at k20 mC until used.

Bone marrow donor. The bone marrow was obtained from a 60year-old male deferred blood donor, deferred when found HCV positive at the implementation of blood donor screening for anti-HCV. He had no symptoms of the infection, and the time and source of infection are unknown. He had received no antiviral treatment for the infection. He had been viraemic for several years as determined by retrospective analyses on stored sera. The HCV genotype in serum at the time of bone

Expression of Fab and IgG in CHO cells. Seven clones (L1, L3, 1 : 5, 1 : 7, 1 : 11, A8 and A12) were expressed as Fab fragments and whole IgG1 in CHO cells. The genes encoding heavy and light chains for each clone were transferred from pComb3H to the vectors pcLCHC and pclgG1, transfected into CHO cells by using Lipofectamine (Life Technologies) according to the manufacturer’s instructions, and expressed as detailed by Samuelsson et al. (1996). Antibodies were

Methods

CEFC

Human monoclonal antibodies to HCV secreted into the medium, and cell culture supernatant was collected and stored in aliquots in k20 mC. The unpurified supernatant was used for most of the subsequent assays. However, for epitope mapping studies and the NOB assay, IgG was further purified from the culture supernatant on Protein G columns (Amersham Pharmacia).

added to duplicate wells. After a 2 h incubation at 37 mC and subsequent washes, AP–streptavidin (Pierce) was added for 1 h, the plate washed, and substrate added. The negative control competitor was a human monoclonal IgG1 antibody (Gp13) directed against HIV-1 gp120 (Schutten et al., 1993).

ELISA for quantification of Fab and IgG. The concentrations of expressed Fab and IgG were determined by ELISA as previously described (Samuelsson et al., 1995, 1996).

Inhibition of E2 binding to cells. E. coli-produced Fab proteins, as well as Fab and Protein G-purified IgG1 expressed in eukaryotic cells, were tested for their capacity to block the binding of recombinant E2 to MOLT-4 cells. This was assessed by flow cytometry using the NOBassay as detailed elsewhere by Rosa et al. (1996).

ELISA for anti-HCV E2 and E1/E2 reactivity. Recombinant E2 (genotype 1a or 1b) or the E1\E2 heterodimer (genotype 1a) was diluted to 1 µg\ml in 0n05 M carbonate–bicarbonate buffer, pH 9n6, and coated to microtitre wells (Costar 3690) overnight at 4 mC. Unbound antigen was discarded, and the wells were blocked with 5 % non-fat dry milk in PBS for 60 min at room temperature. Blocking solution was discarded, and antibody solutions to be tested added in 1 : 1, 1 : 10 and 1 : 100 dilutions (diluent : PBS with 0n1 % NP-40). The plates were incubated at 37 mC for 2 h and washed five times with PBS with 0n05 % Tween 20 (PBS-T) ; alkaline phosphatase (AP)-coupled-goat anti-human F(abh) # (Pierce) or AP–goat anti-human IgG (Fc) (ANL produkter, A$ lvsjo$ , Sweden) antibodies in a 1 : 1000 dilution were then added. After 60 min at 37 mC and subsequent washes, substrate solution (p-nitrophenyl phosphate, Sigma) was added and absorbance measured at 405 nm. Cutoff value for positive readings was deemed to be four times the absorbance value obtained for a negative control sample, an anti-HIV Fab or IgG of equal concentrations (Burton et al., 1991 ; Samuelsson et al., 1996). For control purposes, BSA (Sigma), HIV gp120LAI (Intracell, Cambridge, MA, USA) and tetanus toxoid (SBL Vaccin, Solna, Sweden) coated at 1 µg\ml were used in corresponding ELISAs to control for unspecific reactivity.

Sequencing of Fab clones. Sequencing of the Fab clones was performed utilizing fluorescein-labelled primers [SEQKb (5h ATA GAA GTT GTT CAG CAG GCA) and omp-seq (5h AAG ACA GCT ATC GCG ATT GCA G) for κ light chains, and SEQGb (5h GTC GTT GAC CAG GCA GCC CAG) and pel-seq (5h ACC TAT TGC CTA CGG CAG CCG) for γ heavy chains] and an ALF automated sequencer (Pharmacia Biotech).

Inhibition ELISA for affinity determination. The affinity of the Fab and IgG molecules for the E2 antigens was estimated using an inhibition ELISA (Persson et al., 1991 ; Rath et al., 1988). Samples to be tested were first titrated at 10-fold dilutions in order to find a suitable concentration from which 10-fold reduced concentration gave a substantial reduction in detected binding in the anti-E2 ELISA (as above). Coating of microtitre wells with E2 antigen and subsequent blocking were done as described for the anti-E2 ELISA. Fab and IgG samples, at appropriate dilutions, with or without added soluble E2 (at tenfold dilutions, maximum concentration 5 µg\ml) were added to the wells, and incubated at 37 mC for 2 h. The plates were washed uniformly four times with PBS-T, and bound antibody was detected using AP–anti-Fab or AP–anti-Fc, as described above for ELISA. The reduction in absorbance in the presence of soluble E2 was calculated and the concentration needed for a 50 % reduction estimated by extrapolation.

Cross-competition ELISA. A competitive ELISA was used to determine whether the cloned antibodies shared the same epitopes. Recombinant E1\E2 (genotype 1a) protein was used as antigen and experimental conditions were as described above for the anti-E2 ELISA. Clone L1 IgG was biotinylated using EZ-Link Biotin-LC-Hydrazide (Pierce) according to the manufacturer’s instructions. Equal volumes of the biotinylated L1-IgG at approximately 0n06 µg\ml and IgG of one of the other clones, at approximately 0n1 and 0n025 µg\ml, were mixed and

Results Library size

The ligation of κ light chain genes into pComb3H gave a library of 2i10( members. The subsequent ligation of γ1 Fd genes into the light chain library resulted in a combinatorial library with 2i10' members. Isolation of E2 reactive clones

After selection by panning, clones were initially screened for E2 reactivity and promptly sequenced, in order to identify multiple copies of the same original clone. In panning series I, a 100-fold increase in eluted phage was noted after three rounds of selection. Ten clones were assayed (L1–L10), and all were found to be strongly reactive with the E2 antigen with no cross-reactions to non-HCV antigens. Surprisingly, all of them carried very similar CDR3 sequences in their heavy chains (the H3 region), indicating that they were derived from the same Bcell clone (Litwin & Shlomchik, 1990). However, while the VDJ junctions and the length of the H3 region were identical, they did show a number of different point mutations in their heavy chains, and were combined with different light chains (data not shown). To isolate additional clones, another, single panning was performed (series II). From this experiment, of 20 clones initially tested, ten produced sufficient amounts of anti-E2 specific Fab. Sequencing showed that seven carried heavy chains related to the ones found in the series I panning (the L1–L10 clones). Three clones had H3 regions that were distinct from the L-clones (1 : 5, 1 : 7, 1 : 11). However, clones 1 : 5 and 1 : 11 had genetically related H3 regions. A third selection experiment (panning series III) was performed using the E1\E2 heterodimer antigen. No significant difference in eluted phage was observed. After three rounds of panning, 30 clones were assayed, of which 16 were E1\E2 reactive. All 16 clones were also reactive with E2 alone. Twelve clones had an H3 sequence similar to L1–L10, while four of the clones had unique H3 sequences. Two of the clones were selected for further analysis because they had a higher affinity to E2 (clones A8 and A12). In summary, of 50 clones tested from series I–III, 36 were E2 specific, and 29 shared a related heavy chain. The deduced amino acid sequences of the variable regions of seven clones CEFD

T. Allander and others (a)

(b)

Fig. 1. Deduced amino acid sequences of the variable light (a) and heavy (b) chain regions for seven clones (four carrying unique heavy chains). The CDR regions as suggested by Kabat et al. (1991) are underlined.

(L1, L3, 1 : 5, 1 : 7, 1 :11, A8 and A12) are shown in Fig. 1 (a, b). The antibodies reacted with the HCV antigens in ELISA with a higher reactivity to the E1\E2 complex compared to E2 only, CEFE

but not to control antigens (Fig. 2). Clones 1 : 7, A8 and A12 reacted with E2 derived from HCV of both genotype 1a and 1b in ELISA.

ELISA reading (A405)

Human monoclonal antibodies to HCV

Fig. 2. ELISA reactivity for seven Fab clones with recombinant E2 protein of genotype 1a and 1b, E1/E2 protein complex (genotype 1a), and control antigens (BSA and HIV-1 gp120). The bars indicate the mean absorbance at 405 nm.

Clone

No difference in antigen-specific reactivity in ELISA was seen when comparing Fab molecules with the Fd part expressed as a fusion protein to cpIII of phage, or as an ordinary Fab (FdjLC) or when comparing bacterially expressed Fab with Fab or IgG of the same clone expressed in CHO cells (data not shown). Expression levels

In E. coli, most Fab clones were found to produce 0n2–2n0 mg Fab\l of culture, corresponding to 10–100 µg\ml in the periplasmic preparations. In CHO cell culture supernatants, the yield was between 0n1 and 0n5 µg\ml for Fabs and 0n5–1n0 µg\ml for IgG. After Protein G affinity purification of IgG1, concentrations of 10–40 µg\ml were obtained in the preparations.

Table 1. Approximate affinities for recombinant E2 glycoprotein (genotype 1a) of the antibody clones expressed as IgG in CHO cells or Fab in E. coli Affinity values given in M−". Clone

IgG

Fab

L1 L3 1:5 1:7 1 : 11 A8 A12

4i10( 4i10( 1i10( 2i10) 4i10( Not tested Not tested

4i10( 4i10( 1i10( 2i10)* 4i10( 2i10)* 1i10(*

* Affinity of Fab was also tested for E2 of genotype 1b, and results were similar to those obtained with E2 of genotype 1a.

Detection of heavy and light chains by Western blot

To test for correct expression of both chains, the IgG proteins were analysed in Western blot for each of the five clones expressed in CHO cells, using antiserum for human Fcand light chains. Both chains were approximately equally wellexpressed in all clones (data not shown). Furthermore, the expression of Fd- and light chains in four bacterially expressed clones was investigated (L1, L3, A8, A12), and both chains were detected. The Fd-chain expressed in fusion with the truncated phage protein cpIII showed an approximate molecular mass of 70 kDa (data not shown). Reactivity with the E2 protein in Western blot

None of the seven Fab proteins tested reacted with E2 of genotype 1a in the Western blot, but serum from the genotype

2b-infected bone marrow donor, drawn at the same time as the bone marrow aspiration, did react with the denatured E2 protein. However, when the antigen was gel-separated under non-denaturing conditions, the two clones (1 : 7 and A8) tested bound the native E2 protein on the nitrocellulose strip under otherwise identical assay conditions. Affinity for the E2 protein

The approximate affinities of the antibody clones for the recombinant E2 antigen of genotype 1a, as determined in an inhibition ELISA, varied between 1i10( and 2i10) M−" (Table 1). Affinities were the same whether the antibodies were expressed as Fab fragments in E. coli, or Fab fragments or IgG in CHO cells. In addition, the three clones that reacted with the CEFF


Table 2. NOB titres of anti-E2 antibody clones The Fab or IgG concentration (in µg\ml) needed to block the binding of E2 by 50 % is given (NOB ) (Rosa et al., 1996). &! (a) Blocking of E2 glycoprotein (genotype 1a) binding to target cells by purified IgG1 tested for four antibody clones specific for distinct epitopes IgG1 clone Fig. 3. Cross-competition of the isolated antibody clones for binding to recombinant E1/E2 (1a) determined by ELISA. The binding of the biotinylated IgG clone L1 (at 0n06 µg/ml) was inhibited by addition of the other IgG clones [at 0n1 ( ) and 0n025 ( ) µg/ml]. The bars indicate percentage inhibition of absorbance compared to the value obtained for L1 only binding to the antigen. Gp13 was a negative control IgG directed against HIV-1 gp120 (Schutten et al., 1993).

E2 antigen of genotype 1b (1 : 7, A8, A12) had equal affinities for both genotypes. Epitope mapping

Cross-competition findings for the seven clones investigated indicated that five clones competed for the same or overlapping epitopes as clone L1 (Fig. 3). Clone 1 : 7 did not compete for the ‘ L1-epitope ’, and clone A12 showed only a slight inhibition. These two clones must be assumed to bind to other parts of E2. Inhibition of E2 binding to cells

Clones representing binding to distinct epitopes were tested as purified IgG1 for inhibition of E2 binding to target cells in the NOB assay (Rosa et al., 1996). All clones (L3, 1 : 7, A8, A12) tested efficiently inhibited the binding of recombinant E2 glycoprotein (type 1a) to MOLT-4 cells at a 50 % inhibition concentration of 0n3–0n5 µg\ml. A human monoclonal IgG1 antibody (CL3, expressed and purified in the same manner) to HIV-1 envelope glycoprotein gp120 was negative (Table 2 a). All seven clones characterized were also tested for blocking of E2 (both genotype 1a and 1b) as E. coli-expressed, nonpurified periplasmic preparations of Fabs (clones L1, L3, 1 : 5, 1 : 7, 1 : 11, A8 and A12). They were all positive for NOB activity, even at a 10–100-fold lower concentration than found for the purified IgG1 ; clones A8, 1 :7, L1 and L3 had very high NOB activities (Table 2 b). Two control anti-HIV gp120 Fab clones had no NOB activity (Table 2 b).

Discussion In the present report we describe the isolation of 36 E2specific antibody clones from a patient chronically infected by HCV. The choice of a chronically infected donor, who is obviously not protected from HCV infection, for this type of CEFG

1:7 L3 A8 A12 CL3*

NOB50 (µg/ml) 0n5 0n3 0n4 0n3 10

* CL3 is an anti-HIV-1 directed human monoclonal antibody (IgG1) (Samuelsson et al., 1996).

(b) Fab clones tested as non-purified periplasmic preparations, with E2 antigen of both genotype 1a and 1b NOB50 (µg/ml) Fab clone 1:5 1:7 1 : 11 L1 L3 A8 A12 b12* b14*

E2 1a

E2 1b

2n5 0n01 0n1 0n03 0n02 0n001 0n1 10 10

2n5 0n02 0n15 0n03 0n03 0n007 0n15 10 10

* b12 and b14 are anti-HIV-1 directed human monoclonal antibodies (Fab, non-purified periplasmic preparations (Burton et al., 1991).

experiment may seem irrational. However, experimental evidence suggests that protective antibodies are present during chronic infection (Farci et al., 1994), and indeed, generation of antibodies to HIV-1 using exactly the same approach has resulted in antibody clones having excellent neutralizing activity, including protection from infection in an animal model (Burton et al., 1994 ; Gauduin et al., 1997). The donor was infected with HCV genotype 2b, and the antigens used for selection were genotype 1a. This approach selects for antibodies directed to conserved epitopes which are supposedly of greater interest for development of vaccine and immunotherapy (Burioni et al., 1998). On the other hand, antibodies directed to genotype-specific or strain-specific epitopes will be lost. Whether strain-specific antibodies are necessary for protection is unknown. Some reports suggest that this may be

Human monoclonal antibodies to HCV

the case (Farci et al., 1994), while others point in the opposite direction (Feray et al., 1998 ; Ishii et al., 1998 ; Rosa et al., 1996 ; Scarselli et al., 1995). Twenty-nine of the clones shared related heavy chains, indicating that they represent different stages of maturation of a single original B-cell clone. The striking dominance of this specificity may be due to a methodological bias, e.g. local predominance of a population of related B-lymphocytes at the site of bone marrow aspiration, or efficient maintenance of this particular sequence in E. coli. However, it is also possible that the majority of serum antibodies to E2 share these characteristics. Thus, the epitope to which they are directed may well be an immunodominant epitope. Seven antibody clones representing different heavy chains were analysed in detail. Five of the clones were shown to bind the same or overlapping epitopes, while two clones (A12 and 1 : 7) bound to one or two other epitopes on E2. None of the antibody clones recognized denatured E2 in a Western blot assay, strongly indicating that they recognize conformationdependent epitopes. Such antibodies may be useful for assessing the conformational states of E2 during intracellular processing and virus assembly (Deleersnyder et al., 1997). Indeed, we noted higher ELISA reactivities for the antibodies to the E1\E2 complex compared to E2 with the same sequence. The E1\E2 complex is thought to contain the native form of E2 assembled on the virion, but only a few monoclonal antibodies distinguish well-folded E2 in heterodimers from other forms of E2 (Deleersnyder et al., 1997). To assess whether our antibodies may be used in this regard requires further studies. It is possible that differences in concentration of antigen account for the observed differences of reactivity. Virus neutralizing antibodies are of particular interest for developing immunotherapy and as underlying bases for vaccine design (Chanock et al., 1993). In the case of HCV, assessing the possible neutralizing capacity of antibodies is difficult in the absence of a reliable in vitro neutralization assay. The NOB assay, measuring antibody inhibition of recombinant E2 binding to CD81 on target cells, may have relevance for neutralization although the extent to which CD81 is critical for the cell–HCV interaction has not been fully elucidated. Still, available data suggest that antibodies blocking this interaction may be protective in vivo (Ishii et al., 1998 ; Rosa et al., 1996). Four of our clones, representing specificities with distinct epitopes, were tested as purified, eukaryotic-expressed IgG1 for NOB activity : they were positive at concentrations of 0n3–0n5 µg\ml. Crude preparations of Fab molecules produced in E. coli showed higher NOB titres (0n001–2n5 µg\ml) ; the negative control Fab preparations remained negative. This discrepancy between NOB activity for the same clone if produced as Fab in bacteria or as IgG in CHO cells may have several explanations. First of all, the Fab preparations contain large amounts of bacterial substances, some of which may well have affinity for Ig-related structures. Second, Fab molecules expressed in bacteria have a tendency to aggregate. Both these

factors may bias the ELISA used for quantification of Fab, and\or the cell-based NOB assay (including the secondary antibody reagents used in the latter). This problem is usually addressed by purifying the bacterial Fab. In the present study we instead moved the antibody genes to a different vector for expression in another host before purification of whole IgG. For some clones, the expression of eukaryotic genes in a eukaryotic host has been shown to provide antibody molecules of better efficiency, and is therefore to us more attractive (Samuelsson et al., 1996). In addition, it rules out the possibility that any remaining bacterial contaminants may produce artefactual results. Accordingly, we consider that data obtained with eukaryotic cell-produced IgG are more reliable for calculation of NOB activity, and conclude that the clones isolated are positive for blocking the E2–CD81 interaction at an approximate concentration of 0n5 µg\ml, a NOB titre in between those previously published for human anti-E2 antibodies (Fab and IgG) (Burioni et al., 1998 ; Habersetzer et al., 1998). We must also conclude that the NOB assay is not suitable for titrating crude periplasmic preparations of bacterially expressed Fabs. However, since the anti-HIV control Fab remained negative, the assay probably discriminates E2-binding Fabs from non-binding Fabs. Thus, all anti-E2 Fabs tested may actually bind E2 of both genotypes 1a and 1b, although some of the clones bound poorly to E2 genotype 1b coated to the solid phase in ELISA. This may suggest that the antibodies recognize only the small proportion of biologically active, well-folded E2 molecules – those active in CD81 binding (Flint et al., 2000). Thus, the antibody clones were reactive with E2 of genotype 1a and 1b. The antibodies were very likely elicited by HCV type 2b (there was no E2 of genotype 2 available to us for testing the binding to this particular genotype). Taken together, these data suggest the existence of two or three cross-genotype conserved epitopes in HCV E2 that are conformation dependent, and are targets of antibody-mediated inhibition of the E2–CD81 interaction ; as discussed above, such antibodies may be protective in vivo (Ishii et al., 1998 ; Rosa et al., 1996). The use of monoclonal antibodies may help to better resolve the issue of protective antibodies to HCV. It is likely that natural infection induces only low titres of neutralizing antibodies, and may even result in mostly non-neutralizing interfering antibodies, as has been reported for HIV (Parren et al., 1997). Thus, serum from infected individuals may not be a good reagent for assessing protective antibodies to HCV, and the confusing results obtained so far must be interpreted with this in mind (Farci et al., 1992, 1994 ; Krawczynski et al., 1996). The present study represents one approach to overcome this problem. In conclusion, we have isolated NOB-positive human monoclonal antibodies to HCV E2 that are reactive to two or three different conformational epitopes that seem to be conserved between different genotypes. Further assessment of CEFH


their antiviral properties has to await a reliable in vitro neutralization assay. Broadly reactive anti-E2 antibodies may be valuable for therapeutic applications and provide valuable knowledge for vaccine development. The expert technical assistance of Eva von Garrelts and Fayezeh Yari is gratefully acknowledged. The bone marrow aspirate was kindly provided by Dr Ewa Wallmark. The study was approved by the Regional Ethics Committee of the Karolinska Institute (no. 90 : 261), Stockholm, and written approved consent was obtained from the bone marrow donor after he had read information approved by the Ethics Committee about the project and the procedures. This work was supported by grants from the Swedish Cancer Society, the Swedish Research Council for Engineering and Technological Sciences, the Swedish Medical Research Council, the Torsten and Ragnar So$ derberg Foundations, the Nanna Svartz Fund, the Swedish Society of Medicine, ‘ Fo$ renade Liv ’ Mutual Group Life Insurance Company, the Funds of the Swedish Royal Academy of Sciences, and the Funds of the Karolinska Institute.

References Allander, T., Beyene, A., Jacobson, S. H., Grillner, L. & Persson, M. A. A. (1997). Patients infected with the same hepatitis C virus strain display

different kinetics of the isolate-specific antibody response. Journal of Infectious Diseases 175, 26–31. Alter, H. J., Purcell, R. H., Shih, J. W., Melpolder, J. C., Houghton, M., Choo, Q. L. & Kuo, G. (1989). Detection of antibody to hepatitis C virus

in prospectively followed transfusion recipients with acute and chronic non-A, non-B hepatitis. New England Journal of Medicine 321, 1494–1500. Barbas, C. F., III & Wagner, J. (1995). Synthetic human antibodies : selecting and evolving functional proteins. Methods 8, 94–103. Barbas, C. F., III, Kang, A. S., Lerner, R. A. & Benkovic, S. J. (1991).

Assembly of combinatorial antibody libraries on phase surfaces : the gene III site. Proceedings of the National Academy of Sciences, USA 88, 7978–7982. Bjøro, K., Froland, S. S., Yun, Z., Samdal, H. H. & Haaland, T. (1994).

Hepatitis C infection in patients with primary hypogammaglobulinemia after treatment with contaminated immune globulin. New England Journal of Medicine 331, 1607–1611. Burioni, R., Plaisant, P., Manzin, A., Rosa, D., Delli Carri, V., Bugli, F., Solforosi, L., Abrignani, S., Varaldo, P. E., Fadda, G. & Clementi, M. (1998). Dissection of human humoral immune response against hepatitis

C virus E2 glycoprotein by repertoire cloning and generation of recombinant Fab fragments. Hepatology 28, 810–814. Burton, D. R., Barbas, C. F., III, Persson, M. A. A., Koenig, S., Chanock, R. M. & Lerner, R. A. (1991). A large array of human monoclonal

antibodies to type 1 human immunodeficiency virus from combinatorial libraries of asymptomatic seropositive individuals. Proceedings of the National Academy of Sciences, USA 88, 10134–10137. Burton, D. R., Pyati, J., Koduri, R., Sharp, S. J., Thornton, G. B., Parren, P. W., Sawyer, L. S., Hendry, R. M., Dunlop, N., Nara, P. L. and others

(1994). Efficient neutralization of primary isolates of HIV-1 by a recombinant human monoclonal antibody. Science 266, 1024–1027. Chanock, R. M., Crowe, J. E., Jr, Murphy, B. R. & Burton, D. R. (1993).

Human monoclonal antibody Fab fragments cloned from combinatorial libraries : potential usefulness in prevention and\or treatment of major human viral diseases. Infectious Agents and Disease 2, 118–131. Chomczynski, P. & Sacchi, N. (1987). Single-step method of RNA isolation by acid guanidinium thiocyanate–phenol–chloroform extraction. Analytical Biochemistry 162, 156–159. CEFI

Deleersnyder, V., Pillez, A., Wychowski, C., Blight, K., Xu, J., Hahn, Y. S., Rice, C. M. & Dubuisson, J. (1997). Formation of native hepatitis

C virus glycoprotein complexes. Journal of Virology 71, 697–704. Farci, P., Alter, H. J., Govindarajan, S., Wong, D. C., Engle, R., Lesniewski, R. R., Mushahwar, I. K., Desai, S. M., Miller, R. H., Ogata, N. and others (1992). Lack of protective immunity against reinfection

with hepatitis C virus. Science 258, 135–140. Farci, P., Alter, H. J., Wong, D. C., Miller, R. H., Govindarajan, S., Engle, R., Shapiro, M. & Purcell, R. H. (1994). Prevention of hepatitis

C virus infection in chimpanzees after antibody-mediated in vitro neutralization. Proceedings of the National Academy of Sciences, USA 91, 7792–7796. Farci, P., Shimoda, A., Wong, D., Cabezon, T., De Gioannis, D., Strazzera, A., Shimizu, Y., Shapiro, M., Alter, H. J. & Purcell, R. H. (1996). Prevention of hepatitis C virus infection in chimpanzees by

hyperimmune serum against the hypervariable region 1 of the envelope 2 protein. Proceedings of the National Academy of Sciences, USA 93, 15394–15399. Feray, C., Gigou, M., Samuel, D., Ducot, B., Maisonneuve, P., Reynes, M., Bismuth, A. & Bismuth, H. (1998). Incidence of hepatitis C in

patients receiving different preparations of hepatitis B immunoglobulins after liver transplantation. Annals of Internal Medicine 128, 810–816. Flint, M., Dubuisson, J., Maidens, C., Harrop, R., Guile, G. R., Borrow, P. & McKeating, J. A. (2000). Functional characterization of intracellular

and secreted forms of a truncated hepatitis C virus E2 glycoprotein. Journal of Virology 74, 702–709. Gauduin, M. C., Parren, P. W., Weir, R., Barbas, C. F., Burton, D. R. & Koup, R. A. (1997). Passive immunization with a human monoclonal

antibody protects hu-PBL-SCID mice against challenge by primary isolates of HIV-1. Nature Medicine 3, 1389–1393. Habersetzer, F., Fournillier, A., Dubuisson, J., Rosa, D., Abrignani, S., Wychowski, C., Nakano, I., Trepo, C., Desgranges, C. & Inchauspe, G. (1998). Characterization of human monoclonal antibodies specific to the

hepatitis C virus glycoprotein E2 with in vitro binding neutralization properties. Virology 249, 32–41. Ishii, K., Rosa, D., Watanabe, Y., Katayama, T., Harada, H., Wyatt, C., Kiyosawa, K., Aizaki, H., Matsuura, Y., Houghton, M., Abrignani, S. & Miyamura, T. (1998). High titers of antibodies inhibiting the binding of

envelope to human cells correlate with natural resolution of chronic hepatitis C. Hepatology 28, 1117–1120. Kabat, E. A., Wu, T. T., Perry, H. M., Gottesman, K. S. & Foeller, C. (1991). Sequences of Proteins of Immunological Interest, 5th edn. Bethesda,

MD : U.S. Department of Health and Human Services. Kang, A. S., Burton, D. R. & Lerner, R. A. (1991). Combinatorial immunoglobulin libraries in phage lambda. Methods : A Companion to Methods in Enzymology 2, 1111–1118. Kato, N., Sekiya, H., Ootsuyama, Y., Nakazawa, T., Hijikata, M., Ohkoshi, S. & Shimotohno, K. (1993). Humoral immune response to

hypervariable region 1 of the putative envelope glycoprotein (gp70) of hepatitis C virus. Journal of Virology 67, 3923–3930. Krawczynski, K., Alter, M. J., Tankersley, D. L., Beach, M., Robertson, B. H., Lambert, S., Kuo, G., Spelbring, J. E., Meeks, E., Sinha, S. & Carson, D. A. (1996). Effect of immune globulin on the prevention of

experimental hepatitis C virus infection. Journal of Infectious Diseases 173, 822–828. Litwin, S. & Shlomchik, M. (1990). A test for clonal relatedness in a set of lymphocytes. Journal of Experimental Medicine 171, 293–297. Parren, P. W., Burton, D. R. & Sattentau, Q. J. (1997). HIV-1 antibody – debris or virion ? Nature Medicine 3, 366–367.

Human monoclonal antibodies to HCV Persson, M. A. A., Caothien, R. H. & Burton, D. R. (1991). Generation

of diverse high-affinity human monoclonal antibodies by repertoire cloning. Proceedings of the National Academy of Sciences, USA 88, 2432–2436. Pileri, P., Uematsu, Y., Campagnoli, S., Galli, G., Falugi, F., Petracca, R., Weiner, A. J., Houghton, M., Rosa, D., Grandi, G. & Abrignani, S. (1998). Binding of hepatitis C virus to CD81. Science 282, 938–941. Rath, S., Stanley, C. M. & Steward, M. W. (1988). An inhibition enzyme

immunoassay for estimating relative antibody affinity and affinity heterogeneity. Journal of Immunological Methods 106, 245–249. Rehermann, B., Chang, K. M., McHutchinson, J., Kokka, R., Houghton, M., Rice, C. M. & Chisari, F. V. (1996). Differential cytotoxic T-

lymphocyte responsiveness to the hepatitis B and C viruses in chronically infected patients. Journal of Virology 70, 7092–7102. Reichard, O., Norkrans, G., Fryden, A., Braconier, J. H., So$ nnerborg, A. & Weiland, O. (1998). Randomised, double-blind, placebo-controlled

trial of interferon alpha-2b with and without ribavirin for chronic hepatitis C. The Swedish Study Group. Lancet 351, 83–87. Rosa, D., Campagnoli, S., Moretto, C., Guenzi, E., Cousens, L., Chin, M., Dong, C., Weiner, A. J., Lau, J. Y., Choo, Q. L., Chien, D., Pileri, P., Houghton, M. & Abrignani, S. (1996). A quantitative test to estimate

neutralizing antibodies to the hepatitis C virus : cytofluorimetric assessment of envelope glycoprotein 2 binding to target cells. Proceedings of the National Academy of Sciences, USA 93, 1759–1763. Samuelsson, A., Chiodi, F., O$ hman, P., Putkonen, P., Norrby, E. & Persson, M. A. A. (1995). Chimeric macaque\human Fab molecules neutralize simian immunodeficiency virus. Virology 207, 495–502. Samuelsson, A., Yari, F., Hinkula, J., Ersoy, O., Norrby, E. & Persson, M. A. A. (1996). Human antibodies from phage libraries : neutralizing

activity against human immunodeficiency virus type 1 equally improved after expression as Fab and IgG in mammalian cells. European Journal of Immunology 26, 3029–3034. Scarselli, E., Cerino, A., Esposito, G., Silini, E., Mondelli, M. U. & Traboni, C. (1995). Occurrence of antibodies reactive with more than

one variant of the putative envelope glycoprotein (gp70) hypervariable region 1 in viremic hepatitis C virus-infected patients. Journal of Virology 69, 4407–4412. Schutten, M., McKnight, A., Huisman, R. C., Thali, M., McKeating, J. A., Sodroski, J., Goudsmit, J. & Osterhaus, A. D. (1993). Further

characterization of an antigenic site of HIV-1 gp120 recognized by virus neutralizing human monoclonal antibodies. AIDS 7, 919–923. Shimizu, Y. K., Hijikata, M., Iwamoto, A., Alter, H. J., Purcell, R. H. & Yoshikura, H. (1994). Neutralizing antibodies against hepatitis C virus

and the emergence of neutralization escape mutant viruses. Journal of Virology 68, 1494–1500. Shimizu, Y. K., Igarashi, H., Kiyohara, T., Cabezon, T., Farci, P., Purcell, R. H. & Yoshikura, H. (1996). A hyperimmune serum against a

synthetic peptide corresponding to the hypervariable region 1 of hepatitis C virus can prevent viral infection in cell cultures. Virology 223, 409–412. Spaete, R. R., Alexander, D., Rugroden, M. E., Choo, Q. L., Berger, K., Crawford, K., Kuo, C., Leng, S., Lee, C., Ralston, R. and others (1992).

Characterization of the hepatitis C virus E2\NS1 gene product expressed mammalian cells. Virology 188, 819–830. Widell, A., Shev, S., Ma/ nsson, S., Zhang, Y. Y., Foberg, U., Norkrans, G., Fryden, A., Weiland, O., Kurkus, J. & Nordenfelt, E. (1994).

Genotyping of hepatitis C virus isolates by a modified polymerase chain reaction assay using type specific primers : epidemiological applications. Journal of Medical Virology 44, 272–279. Zibert, A., Schreier, E. & Roggendorf, M. (1995). Antibodies in human sera specific to hypervariable region 1 of hepatitis C virus can block viral attachment. Virology 208, 653–661. Zibert, A., Meisel, H., Kraas, W., Schulz, A., Jung, G. & Roggendorf, M. (1997). Early antibody response against hypervariable region 1 is

associated with acute self-limiting infections of hepatitis C virus. Hepatology 25, 1245–1249. Received 14 April 2000 ; Accepted 7 July 2000

CEFJ