Cytomegalovirus (CMV) is a (usually non-pathogenic) virus commonly found in the human population. Following infection, the virus persists in the host but the carrier is protected from disease by its immune system. Both cellular and humoral (antibody) activities are believed to be involved in the defense against such CMV-induced active disease. The Department of Immunotechnology has had a longstanding interest in research on antibody responses in humans against cytomegalovirus. The research has mainly focused on such antibody responses against the diagnostically important antigen pp65 (lower matrix protein; ppUL83) and the therapeutically interesting surface glycoprotein B (gB; gp58/116; gpUL55).
Antibodies against CMV gB has long been known to neutralize virus infection. In collaboration with Prof. Radak we have demonstrated that neutralizing antibodies recognizing AD-1 and AD-2 mediate this effect through different mechanisms. Briefly, an antibody against AD-2 prevents membrane fusion down-stream of virus adsorption to cells. However, even at concentrations that completely neutralize virus infection and membrane fusion, ITC88 (a human AD-2-specific monoclonal antibody), just as AD-1 specific human antibodies, does not prevent virus adsorption. However, in contrast to anti-AD-1, the AD-2-specific antibody substantially reduced heparin-resistant cellular receptor binding of CMV. Thus, although both antibody types act down-stream of heparan sulfate proteoglycans binding, they differ in their direct mechanism (Gicklhorn et al., 2003).
|Antibodies against AD-2 (ITC88) as well as AD-1 (ITC52) are able to block fusion of CMV-infected cells with other cells.This effect cannot be mediated by a negative control human monoclonal antibody (MO58) against the pp65 tegument protein. Adapted from Gicklhorn et al. (2003).||Antibodies against CMV gB AD-1 (ITC52) and AD-2 (ITC88) cannot prevent CMV adsorption to cells to any substantial degree, while heparin effectively prevent virus adsorption to heparan proteoglycans. The negative control antibody is MO58 (anti-CMV pp65). Adapted from Gicklhorn et al. (2003).|
|The ITC88 antibody against AD-2 is able to reduce heparin-resistant binding of CMV to cells, an effect that is not seen with anti-AD-1 (ITC52) or the negative control antibody MO58 (anti-pp65). Adapted from Gicklhorn et al. (2003).|
It has been shown that the human antibody response to the major neutralization determinant AD-1 residing in gp58 is highly diverse. Several overlapping epitopes can be found within the AD-1 structure and each epiope (defined as reactivity against specific mutants and protein fragments) may bind both neutralizing and non-neutralizing antibodies (Schoppel et al., 1996; Speckner et al., 1999). The neutralization determinant AD-2 expressed by the gp116 fragment of gB is also diverse and different antibodies against this epitope have different ultra-fine specificity for the epitope (Lantto et al., 2002a; Ohlin et al., 1996; Silvestri et al., 1993). Overall, very little information on how different antibodies neutralize this virus has been available, largely due to the scarcity of different antibodies against this epitope. We have as outlined above recently shown that the repertoire of antibody genes that is capable of actually creating a AD-2 specific paratope is restricted, a factor which may contribute to the relatively low immunogenicity of this potent virus-neutralization epitope (Lantto et al., 2002a). We have now also demonstrated that a divalent antibody format was required for virus neutralization despite a very slow dissociation of monovalent antibody fragments (scFv) from gB. Thus, size and/or cross-linking is a major factor determining neutralization efficacy of antibodies directed towards the AD-2 epitope (Lantto et al., 2002b). Furthermore, dissociation rate of the anti-AD2/AD2 complex seems to correlate with efficacy of neutralization. Importantly, it was also shown that certain clinical isolates carrying modifications within the AD-2 epitope may escape neutralization by some members of an AD-2 specific immune repertoire despite the fact that they still bind these antibody fragments (Lantto et al., 2003) suggesting that the interplay between the humoral immune response and the virus is extremely complex.
|Dissociation of monomeric and dimeric scFv from 15-mer peptide antigen mimicking AD-2 (A and B, respectively) and recombinant gB (C and D, respectively) immobilized onto a Biacore CM-sensorchip.||Neutralization of CMV strain AD169 infection of human embryonic lung fibroblasts in the presence of monomeric and dimeric scFv as well as monomeric scFv dimerized with anti-FLAG M2 antibody (Lantto et al. (2002b)).|
|Comparison of the neutralization of HCMV strain AD169 and the the half-life of the complex between recombinant gB and monomeric AD-2-specific scFv as determined by Biacore analysis (Lantto et al. (2003)).|
When shuffling the light chain of the AD-2 specific antibody ITC88 with a light chain library derived from peripheral blood B cells, it was apparent that this antibody only could use a very restricted group of light chain sequences closely related to the sequence used by the original hybridoma. It is thus suggested that in vivo developed specific antibody VH sequences may display very limited light chain promiscuity. Furthermore, when introducing these light chain sequences the ultra-fine specificity against the AD-2 epitope was sometimes slightly modified without introducing multireactive properties to this antibody. The chain shuffling approach may thus develop new ultrafine specificities not previously found among available monoclonal antibodies but which at least in this case had been observed in polyclonal antibody preparations. These results define the basis for the light chain restriction in this antibody.
In an extension to these studies, the criteria for forming an AD-2 specific paratope in an antibody was further defined. Importantly we have been able to demonstrate the features of a high affinity repertoire against the potently neutralising determinant AD-2 on gB and described possible reasons for the poor immunogenicity of this structure. Mutations of residues in CDRH1 and CDRL1 were critical determinants in allowing an AD-2 specific antibody to gain a high affinity character in its recognition of intact CMV gB.
|Dissociation of an affinity matured, monovalent anti-gB single chain antibody fragment (B) and a sequence variant (A) carrying the germline encoded alanine in position 34H in the heavy chain variable domain from glycoprotein B. Note the importance of the aspartate 34H mutation for efficient binding of the anti-gB antibody to the AD-2 epitope of CMV gB. Dissociation of the antigen-antibody complex was determined by using Biacore technology.|
The sequences of variable region sequences from the early studies are available in GenBank under accession codes: L37300-L37309. Additional sequences are available using accession codes AF443427-AF443433 and AF492001.
We have developed a range of recombinant antibodies against glycoprotein B epitopes AD-1 and AD-2 with different levels of complement activating potential. These antibodies, which carried IgG1, IgG3 and IgG3mA (a synthetic version of IgG3 carrying a shorter and less flexible hinge), were evaluated for their potential to use complement to neutralise virus infection. The following results were obtained:
|Neutralisation of CMV infection using recombinant versions of the ITC52 (anti-AD-1) and ITC88 (anti-AD2) human antibodies. While the neutralisation potential of ITC88 was unaffected by the isotype and the presence of complement, the anti-AD-1 antibody was greatly affected in its biological activity. The modified hinge in IgG3mA provided the best capacity to neutralise CMV in the presence of complement (filled bars). In contrast IgG3 was substantially inferior to IgG1 in mediated complement-independent neutralisation (open bars).|
Antibodies against the major CMV components pp65 and gB are highly diverse in their sequences and they use a wide range of germline sequences. It has, however, been observed that there appears to be an overrepresentation of VH5 segment usage among antibodies specific for the AD-1 epitope of gB. All these antibodies seem to be mutated in comparison to germline sequences suggesting that these immune responses are T cell dependent. Although these antibodies in some cases uses germline genes sometimes associated with auto- or multireactive properties, no such activity has been demonstrated. Furthermore, these CMV-specific antibodies do not express substantial levels of idiotypes frequently associated with these germline genes and such auto-/multireactive antibodies.
The sequences of these variable region sequences are available in GenBank under accession codes: L26531-L26540, L26890-L26896, L26898 and L37310-L37311
During the coarse of these investigations, we have developed a number of specific, high affinity, human monoclonal antibodies recognizing cytomegalovirus pp65 and gB. In addition one human antibody recognizing pp28 has been obtained. These reagents are available for non-commercial purposes to researchers on a collaborative basis upon request.
Since the 1990-ies our research has utilized recombinant DNA technology to investigate and develop human antibodies specific for CMV. Consequently, a large panel of variants of some of the antibodies described below are available. Furthermore, to faciitate the development of neutralizing antibodies specific for CMV we have established and validated a semi-high throughput system for screening specific binders developed through combinatorial antibody library technology (Carlsson et al., 2012).
|CMV gB AD-1||ITC33|
|CMV gB AD-2||ITC88 and variants thereof|
Dr. Mats Ohlin
Dept. of Immunotechnology
Medicon Village (Building 406)
S-223 81 LUND
telephone: +46-46-222 4322
Page Manager: Mats Ohlin
Last updated: 2013-05-14
Phage display: a short introduction