Multimeric humanized varicella-zoster virus antibody ... - CiteSeerX

Journal of General Virology (2001), 82, 1959–1963. Printed in Great Britain ..........................................................................................................................................................................................................

SHORT COMMUNICATION

Multimeric humanized varicella-zoster virus antibody fragments to gH neutralize virus while monomeric fragments do not P. D. Drew,1 M. T. Moss,3† T. J. Pasieka,2 C. Grose,2 W. J. Harris1, 3 and A. J. R. Porter1 1

University of Aberdeen, Department of Molecular and Cell Biology, Institute of Medical Sciences, Foresterhill, Aberdeen AB25 2ZD, UK Department of Microbiology, University of Iowa College of Medicine, Iowa City, IA 52242, USA 3 Scotgen Biopharmaceuticals Inc., Kettock Lodge, Aberdeen Science and Technology Park, Aberdeen AB22 8GU, UK 2

Murine monoclonal antibody 206 (MAb mu206) binds to gH, the varicella-zoster virus (VZV) fusogen, neutralizing the virus in vitro in the absence of complement and inhibiting cell-to-cell spread and egress of VZV in cultured cells. We have humanized this antibody to generate MAb hu206 by complementarity determining region grafting. MAb hu206 retained binding and in vitro neutralizing activity, as well as cross-reactivity with ten different VZV strains. Single-chain antibody fragments (scAb) derived from MAb hu206 were produced in Escherichia coli. These scAb retained the binding properties of the whole antibody. However, monomeric scAb exhibited markedly reduced neutralizing activity compared to the bivalent parental MAb hu206. Shortening the peptide linker joining the VH to the Vκ domain from 14 to 5 or even 0 residues encouraged multimerization and increased neutralizing efficacy. The fact that Fab fragments enzymatically generated from whole MAb hu206 lost their neutralizing potency lent support to the proposal that valency is important for VZV neutralization at this epitope.

The relative roles of antibody and cell mediated immunity in the control of varicella-zoster (VZV) infection are still not clearly understood. In an early and oft cited hypothesis, HopeSimpson (1965) speculated that periodic subclinical VZV reactivation played a pivotal role, boosting antibody titres, and thus suppressing the neuronal VZV infection. With an Author for correspondence : Andy Porter. Fax j44 1224 273144. e-mail a.porter!abdn.ac.uk † Present address : University of Aberdeen, Academic Transfusion Medicine Unit, Department of Medicine and Therapeutics, Regional Transfusion Centre, Foresterhill, Aberdeen AB25 2ZW, UK

0001-7645 # 2001 SGM

expanded knowledge of cell mediated immunity, especially the cytotoxic T cell response, an increased emphasis has been placed on cellular mechanisms (Abendroth & Arvin, 2000). Yet, there remain compelling reasons to suspect that antibody can alter the course of VZV infection. In particular, the large body of data from clinical trials with zoster immune globulin (ZIG) clearly established that intramuscular administration of high titred VZV antibody abrogates a nascent primary VZV infection (Zaia et al., 1983). How antibody exerts this effect is not entirely clear, but the most likely explanation relates to an expanded concept of neutralization of inoculum virus before viraemia occurs (Grose et al., 2000). The VZV specific antibody with the highest documented neutralization titre is directed against VZV glycoprotein gH (Grose, 1990). Of particular interest, one of these anti-gH murine monoclonal antibodies (MAb 206) (Montalvo & Grose, 1986) is capable of neutralizing VZV in vitro in the absence of human complement. Furthermore, blockage of viral egress and inhibition of cell-to-cell spread of VZV in cultured cells have also been demonstrated (Rodriguez et al., 1993). Because of the large body of knowledge about this anti-gH monoclonal antibody and its apparent biological relevance, we selected it as a candidate for humanization. Moreover, we investigated whether recombinant antibody fragments could be generated that retain these neutralizing properties. Transfer of the 206 murine CDRs to human frameworks was achieved by oligonucleotide site-directed mutagenesis (Tempest et al., 1991 ; Nakamaye & Eckstein, 1986) to form the hu VH and hu Vκ sequences. The templates for mutagenesis were M13HuVHLYS, which contains VH framework regions from human antibody NEWM and VH CDRs from murine antibody D1.3 cloned into M13VHPCR1, and M13HuVKLYS, which contains Vκ framework regions from human antibody REI and Vκ CDRs from murine antibody D1.3 cloned into M13VKPCR1 (Riechmann et al., 1988). The oligonucleotides used were : VHCDR1, 5h dTGTCT CACCC AGCTC ATCCA GAATC TACTG AAGTC GAATC CAGAC GCTGT GCAGG TCAGG 3h ; VHCDR2, 5h dTTGCT GGTGT CTCTG AGCAT TGTCA CTCTG TCCTT CAGAG ATGGC BJFJ

P. D. Drew and others

Fig. 1. Bacterial expression vectors. All coding regions were cloned (HindIII/EcoRI) into the multiple cloning site of pUC19, downstream of the lac promoter. pPM1-His was derived from a previously described vector (McGregor et al., 1994 ; Molloy et al., 1995). PelB is a leader sequence directing secretion to the E. coli periplasm (Ward et al., 1989). VH (117 amino acid residues) and Vκ (111 residues) are the variable heavy and light regions of the antibody, respectively. A peptide linker of 14 or 5 amino acid residues links the heavy and light chains in all vectors except p0LINK-His. Hu Cκ denotes the human constant light (kappa) domain (107 residues), which was used for detection in ELISA and Western blot. 6-His denotes 6 histidine residues used for purification by immobilized metal chelate affinity chromatography.

ATATA GTTTA TCGTA CTGCT ATCTG GATTA ATTTC TCCAA TCCACTCA 3h ; VHCDR3, 5h dCCTTG GCCCC AGTAG TCCAT AGCAT AGGAC CACCC TCTTG CACAATAA 3h ; VKCDR1, 5h dTACCA GTGCA TATAA CTAGA GCCAG ATGTA CTGAC ACTTT TGCTG GCCCT ACAGG TGATGGT 3h ; VKCDR2, 5h dCACAC CAGAT TCTAG GTTGG ATGCA AGGTAG 3h ; VKCDR3, 5h dGGTCC CTTGG CCGAA CGTGA ATGGA AGCTC CCTAC TGTGC TGGCA GTAGTAG 3h. Use of these oligonucleotides led to a number of additional murine residues being deliberately substituted into the human variable region frameworks. Specifically, the VHCDR1 oligonucleotide incorporated framework alterations at residues 24 (Val changed to Ala : V24A), 27 (Ser to Phe : S27F) and 28 (Thr to Asp : T28D) and the VHCDR2 oligonucleotide incorporated a framework alteration at residue 71 (Val changed to Arg : V71R). Mammalian expression vectors pSVgpt and pSVhyg (Mulligan & Berg, 1981), containing the hu206 VH and Vκ genes, respectively, together with the appropriate constant BJGA

regions and control elements, were used to transfect NS0 cells (European Collection of Animal Cell Cultures, Porton, UK, ECACC no. 85110505). Humanized antibody was purified by protein A chromatography. The biological property of greatest interest is the ability of the anti-gH antibody to neutralize infectious virus in a plaquereduction assay (Grose et al., 1979). Therefore, we sought to determine whether this property was retained in the humanized antibody. An aliquot of antibody ranging from 1n2 to 50 µg in a 1 ml volume was added to 1 ml of virus suspension and the 2 ml mixture was incubated for 1 h prior to inoculation onto a 35 mm MeWo cell monolayer. When the monolayers were examined 3 days later, no more than ten small plaques were visualized in any of the five monolayers inoculated with both virus and humanized antibody in a final concentration of at least 2n5 µg\ml. The virus control exhibited more than 100 plaques. This observation was easily reproducible and suggested that the humanization procedure had been successful because the recombinant antibody retained the same complement independent neutralization property of the murine antibody. The laboratory strain VZV-32 (Grose & Brunell, 1978) is a low passage virus (fewer than 20 passages) isolated from a child with chickenpox. To be assured that there was no variation in susceptibility to neutralization between wild-type VZV strains, isolates from ten other children with chickenpox, collected in the USA between the years 1977–1990, were tested by plaque reduction. Among these ten wild-type isolates, there were no major differences in susceptibility to neutralization compared to strain VZV-32. In general, most VZV isolates were neutralized by 1n2–2n5 µg\ml of humanized antibody 206. In order to evaluate whether the humanized MAb was superior to VZIG, we purchased two vials of VZIG (Massachusetts Public Health Biologic Laboratories, Boston, MA, USA). Each vial was selected from a different lot (MVZIG-44 and MVZIG-46). The VZV neutralization assays were repeated as previously described. The total protein content of VZIG was 165 mg\ml ; the initial plaque-reduction assays were carried out with the same concentrations of antibody protein, i.e. between 1 and 100 µg\ml. No inhibition of plaque formation was observed. Therefore, the amount of antibody was increased by directly adding VZIG at much larger volumes. Only when a final VZIG concentration of 6 mg\ml was attained was there a greater than 80 % reduction in plaque formation. If a level of 2n5 µg\ml (167 nM) of humanized Ab is considered necessary for plaque inhibition, then a comparable level for VZIG is 6000 µg\ml. Thus, the biological activity of the humanized antibody is approximately 2400 times that of the standard VZIG preparation on a mg per mg basis. There is an obvious caveat to this comparison : even though human VZIG is prepared from whole IgG with a high anti-VZV titre, a major portion of the IgG will have specificities other than the VZV gH antigen.

Humanized VZV antibody to gH

(a)

(b)

Fig. 2. Antigen binding cell monolayer ELISA. In (a), parental hu206 MAb (X) was compared to hu206 scAb-14 ($) binding to VZV infected cells. Binding of hu206 MAb (W) and scAb-14 (#) to uninfected cells was also tested. In (b), binding of hu206 scAb-14 ( ), scAb-5 ($) and scAb-0 (X) to infected cells was measured. Similarly, binding of hu206 scAb-14 ( ), scAb-5 (#) and scAb-0 (W) to uninfected cells was also tested. Each point represents a mean of two replicates, and each experiment was performed in duplicate.

To further investigate the nature of the hu206 antibody– antigen interactions and provide additional insight into the mechanism of neutralization, we cloned and expressed singlechain antibody fragments (hu206 scAb). These are similar to scFv (Bird et al., 1988), but in addition to VH and Vκ they also contain the whole human light chain constant region (hu Cκ), and have a molecular mass of approximately 40 kDa (Fig. 1) (McGregor et al., 1994). A pUC19-based periplasmic expression vector, pPM1-His (Molloy et al., 1995), was used for expression in Escherichia coli XL-1 Blue (Stratagene). Three different constructs were prepared and expressed, with linkers of length 14, 5 and 0 amino acid residues (Fig. 1). Antibody fragments were purified by immobilized metal affinity chromatography (Dooley et al., 1998). Protein concentration was determined by capture ELISA (McGregor et al., 1994) and protein samples analysed on 15 % SDS– polyacrylamide gels (Sambrook et al., 1989) and size exclusion HPLC (McGregor et al., 1994) (results not shown). These antibody fragments were initially tested for antigen binding in a MeWo cell monolayer ELISA (Sugano et al., 1991). When compared to the parental MAb (Fig. 2 a), the 14 residue linker hu206 scAb (scAb-14) appears to have similar functionality in binding to infected cells, while neither the MAb nor the scAb appear to bind to uninfected MeWo cells. The 5 amino acid residue linker scAb (scAb-5) appears to have reduced functionality compared to scAb-14 (Fig. 2 b), while the 0 residue linker scAb (scAb-0) exhibited only a small amount of binding activity in this assay. Despite positive results in the cell monolayer ELISA, neutralizing activity could not be detected in the standard hu206 scAb-14 at concentrations up to 2000 nM, while the parental hu206 MAb had a 50 % neutralizing concentration of about 4 nM (Fig. 3). In contrast, scAb-5 and scAb-0, which were predicted to form multimeric diabodies or triabodies, had 50 % neutralizing concentrations of approximately 90 and

Fig. 3. Plaque reduction neutralization assay using antibodies and antibody fragments. The samples tested were hu206 MAb (>), hu206 scAb-14 ($), scAb-5 (X), scAb-0 ( ), scAb-14 pre-incubated with secondary antibody (#) and secondary antibody alone ( ). Each point represents a mean of three replicates, and each experiment was performed in duplicate.

700 nM respectively. To provide further evidence suggesting a role for valency in neutralization, hu206 scAb-14 was preincubated with 50 µg\ml of goat anti-hu Cκ secondary antibody for 1 h at 4 mC, before dilution and use in a neutralizing assay (Fig. 3). When pre-incubated with secondary antibody, which would be expected to induce multimerization, scAb-14 had a 50 % neutralizing concentration of about 20 nM, while neither scAb-14 nor the secondary antibody alone had detectable neutralizing activity. Papain was used to digest whole hu206 antibody (Schofield et al., 1997), and when tested for antigen binding in a cell monolayer ELISA (Fig. 4 a), hu206 Fab had a very similar binding profile to that of the whole hu206 MAb, despite the monomeric nature of the Fab. However, when analysed in a neutralization test (Fig. 4 b), the Fab was unable to reduce the number of plaques by 50 %, even at 500 nM, while the 50 % BJGB

P. D. Drew and others

(a)

(b)

Fig. 4. (a) Antigen binding cell monolayer ELISA of proteolytically generated hu206 Fab fragments and whole MAb hu206. The samples tested were hu206 Fab (X) and MAb ($) binding to VZV infected monolayers, and hu206 Fab (W) and MAb (#) binding to uninfected monolayers. Each point represents a mean of two replicates, and this experiment was performed in duplicate. (b) In vitro neutralization assay of antibodies and proteolytically produced antibody fragments. The samples tested were hu206 Fab ($), hu206 Fab pre-incubated with anti-hu Cκ antibody (X), hu206 MAb (>) and anti-hu Cκ antibody ( ). Each point represents a mean of three replicates, and this experiment was performed in duplicate.

neutralization concentration of the MAb was only 2 nM. When the hu206 Fab was pre-incubated with anti-hu Cκ secondary antibody, as for the scAb, the 50 % neutralization titre was approximately 11 nM (Fig. 4 b). Humanized antibodies have several potential advantages over murine monoclonal antibodies (Winter & Harris, 1993) : (i) their expected reduced immunogenicity allows repeated administration without the induction of a host anti-globulin response, (ii) the human Fc portion allows efficient activation of host effector functions and (iii) the presence of a human Fc prolongs the serum half-life of the antibody. Given the increased in vitro neutralization efficacy of the 206 antibody compared to VZIG, the availability of a humanized version may warrant the re-evaluation of the ability of antibody to modify the progression of established VZV infection in vivo. It appears that single-chain antibody fragments expressed in E. coli can also neutralize VZV, but only when expressed as so-called diabodies or triabodies (Holliger et al., 1993 ; Kortt et al., 1997 ; Lawrence et al., 1998), or if multimerization is induced by a secondary antibody. Similar results were obtained for Fab fragments produced by papain digestion of whole hu206 MAb purified from mammalian cells. These observations raise questions about the mechanism of neutralization of the virus by antibody at this epitope. The apparent requirement for multimerization for neutralization may be explained in various ways. Firstly, it may simply be an issue of size i.e. smaller molecules are less able to block the fusogenic function of gH, due to reduced steric hindrance. The larger MAb, and antibody fragments bound by secondary antibody, may more efficiently neutralize the virus because increased steric hindrance blocks a critical step in the fusion process. Alternatively, they may cause aggregation or cross-linking of virus envelope structures. It is, however, unlikely that the neutralization effect of 206 antibody and fragments is dependent on aggregation of BJGC

the virions, because, despite its name, ‘ cell-free ’ VZV prepared by sonic disruption of infected cells tends to be a heterogeneous mixture of virions associated with fragments of cell membrane (Cohen & Straus, 1996). The literature also suggests that virus aggregation, although possible, is not always the most likely cause of neutralization. Mason et al. (1996) described a single-chain antibody against foot-andmouth disease virus (FMDV) that is only weakly neutralizing in vitro unless incubated with a secondary antibody. Aggregation of the FMDV did not appear to be the mechanism for neutralizing. Thullier et al. (1999) report that a recombinant Fab against dengue virus neutralizes in vitro, but to a degree substantially less than would be expected given its affinity compared to the parental antibody. Like antibody 206, this antibody also binds to an envelope protein, and the authors postulate that the reduced neutralization titre was due to a decrease in steric hindrance caused by the Fab, compared to the MAb. Similarly, Lamarre & Talbot (1995) found that while proteolytically generated Fab fragments, used at sufficiently high concentrations, were able to neutralize coronavirus infectivity both in vitro and in vivo, neutralization was significantly greater in bivalent F(abh) fragments, but this # effect was not mediated by aggregation of virions. The possibility that simply an increased avidity of multimeric antibody (fragments) compared to monomeric fragments is responsible for their increased neutralizing potency cannot be ruled out. Avidity refers to the apparent increase in affinity observed when multiple antigen binding sites for a given epitope are located on a single molecule. The physical basis for this effect is that when the first binding site is bound to a surface, the concentration of additional antigen binding site(s) is dramatically increased. Although any avidity effect due to multimerization forced by shortening the linker failed to manifest in the cell monolayer ELISA, it may be that the

Humanized VZV antibody to gH

distribution or orientation of the 206 epitope is different on cell-free VZV, such that multivalent binding would become possible. We thank Andrew Davison for VZV-Dumas and the MeWo cell line, Phil Cash for helpful advice and use of facilities and Kenny Reay and Phil Carter for DNA sequencing services. This work was partially funded by the Biotechnology and Biological Sciences Research Council, UK, and Scotgen Biopharmaceuticals Inc. Research by C. G. funded by NIH grants AI22795 and AI36884.

References Abendroth, A. & Arvin, A. M. (2000). Host response to primary infection. In Varicella-Zoster Virus, pp. 142–156. Edited by A. M. Arvin & A. A. Gershon. Cambridge : Cambridge University Press. Bird, R. E., Hardman, K. D., Jacobson, J. W., Johnson, S., Kaufman, B. M., Lee, S.-M., Lee, T., Pope, S. H., Riordan, G. S. & Whitlow, M. (1988). Single-chain antigen-binding proteins. Science 242, 423–426. Cohen, J. I. & Straus, S. E. (1996). Varicella-zoster virus and its replication. In Fields Virology, 3rd edn, pp. 2525–2545. Edited by B. N. Fields, D. M. Knipe & P. M. Howley. Philadelphia : Lippincott–Raven. Dooley, H., Grant, S. D., Harris, W. J. & Porter, A. J. (1998).

Stabilization of antibody fragments in adverse environments. Biotechnology and Applied Biochemistry 28, 77–83. Grose, C. (1990). Glycoproteins encoded by varicella-zoster virus : biosynthesis, phosphorylation and intracellular trafficking. Annual Review of Microbiology 44, 59–80. Grose, C. & Brunell, P. A. (1978). Varicella-zoster virus : isolation and propagation in human melanoma cells at 36 and 32 mC. Infection and Immunity 19, 199–203. Grose, C., Edmond, B. J. & Brunell, P. A. (1979). Complementenhanced neutralizing antibody response to varicella-zoster virus. Journal of Infectious Diseases 139, 432–437. Grose, C., Ye, M. & Padilla, J. (2000). Pathogenesis of primary infection. In Varicella-Zoster Virus, pp. 105–122. Edited by A. M. Arvin & A. A. Gershon. Cambridge : Cambridge University Press. Holliger, P., Prospero, T. & Winter, G. (1993). ‘ Diabodies ’ : small bivalent and bispecific antibody fragments. Proceedings of the National Academy of Sciences, USA 90, 6444–6448. Hope-Simpson, R. E. (1965). The nature of herpes zoster : a long-term study and a new hypothesis. Proceedings of the Royal Society of Medicine 58, 9–20. Kortt, A. A., Lah, M., Oddie, G. W., Gruen, C. L., Burns, J. E., Pearce, L. A., Atwell, J. L., McCoy, A. J., Howlett, G. J., Metzger, D. W., Webster, R. G. & Hudson, P. J. (1997). Single-chain Fv fragments of anti-

neuraminidase antibody NC10 containing five- and ten-residue linkers form dimers and with zero-residue linker a trimer. Protein Engineering 10, 423–433. Lamarre, A. & Talbot, P. J. (1995). Protection from lethal coronavirus infection by immunoglobulin fragments. Journal of Immunology 154, 3975–3984.

McGregor, D. P., Molloy, P. E., Cunningham, C. & Harris, W. J. (1994).

Spontaneous assembly of bivalent single chain antibody fragments in Escherichia coli. Molecular Immunology 31, 219–226. Mason, P., Berinstein, A., Baxt, B., Parsells, R., Kang, A. & Rieder, E. (1996). Cloning and expression of a single-chain antibody fragment

specific for foot-and-mouth disease virus. Virology 224, 548–554. Molloy, P., Brydon, L., Porter, A. J. & Harris, W. J. (1995). Separation and concentration of bacteria with immobilized antibody fragments. Journal of Applied Bacteriology 78, 359–365. Montalvo, E. A. & Grose, C. (1986). Neutralizing epitope of varicella zoster virus on native viral glycoprotein gp118 (VZV glycoprotein gpIII). Virology 149, 230–241. Mulligan, R. C. & Berg, P. (1981). Selection for animal cells that express the Escherichia coli gene coding for xanthine–guanine phosphoribosyltransferase. Proceedings of the National Academy of Sciences, USA 78, 2072–2076. Nakamaye, K. L. & Eckstein, F. (1986). Inhibition of restriction endonuclease NciI cleavage by phosphorothioate groups and its application to oligonucleotide-directed mutagenesis. Nucleic Acids Research 14, 9679–9698. Riechmann, L., Clark, M., Waldmann, H. & Winter, G. (1988). Reshaping human antibodies for therapy. Nature 332, 323–327. Rodriguez, J. E., Moninger, T. & Grose, C. (1993). Entry and egress of varicella virus blocked by same anti-gH monoclonal antibody. Virology 196, 840–844. Sambrook, J., Fritsch, E. F. & Maniatis, T. (1989). Molecular Cloning : A Laboratory Manual. Cold Spring Harbor, NY : Cold Spring Harbor Laboratory. Schofield, D. J., Stephenson, J. R. & Dimmock, N. J. (1997). Variations in the neutralizing and haemagglutination-inhibiting activities of five influenza A virus-specific IgGs and their antibody fragments. Journal of General Virology 78, 2431–2439. Sugano, T., Tomiyama, T., Matsumoto, Y. I., Sasaki, S., Kimura, T., Forghani, B., Masuho, Y. & Matsumoto, Y. (1991). A human

monoclonal antibody against varicella-zoster virus glycoprotein III. Journal of General Virology 72, 2065–2073. Tempest, P. R., Bremner, P., Lambert, M., Taylor, G., Furze, J. M., Carr, F. J. & Harris, W. J. (1991). Reshaping a human monoclonal antibody to

inhibit human respiratory syncytial virus infection in vivo. Bio\Technology 9, 266–271. Thullier, P., Lafaye, P., Me! gret, F., Deubel, V., Jouan, A. & Mazie! , J. C. (1999). A recombinant Fab neutralizes dengue virus in vitro. Journal of

Biotechnology 69, 183–190. Ward, E. S., Gussow, D., Griffiths, A. D., Jones, P. T. & Winter, G. (1989). Binding activities of a repertoire of single immunoglobulin

variable domains secreted from Escherichia coli. Nature 341, 544–546. Winter, G. & Harris, W. J. (1993). Humanized antibodies. Immunology Today 14, 243–246. Zaia, J. A., Levin, M. J., Preblud, S. R., Leszczynski, J., Wright, G. G., Ellis, R. J., Curtis, A. C., Valerio, M. A. & LeGore, J. (1983). Evaluation

of varicella-zoster immune globulin : protection of immunosuppressed children after household exposure to varicella. Journal of Infectious Diseases 147, 737–743.

Lawrence, L. J., Kortt, A. A., Iliades, P., Tulloch, P. A. & Hudson, P. J. (1998). Orientation of antigen binding sites in dimeric and trimeric

single chain Fv antibody fragments. FEBS Letters 425, 479–484.

Received 19 January 2001 ; Accepted 30 April 2001

BJGD